RE : samba share on VMS bound volume set
I never tried to use Samba on a volume-set, but I can understand why you encounter the behaviour you describe. In order to be aware of changes (file creations/deletions), Samba uses the internal locks maintained by the file system (XQP). Because those locks are not the same for volume sets, it seems normal that you can't see new or deleted files unless you force Samba to refresh its cache. Honestly, and because the VMS wizqrd discourages the use of volume sets, I don't plan to do anything about that. JYC -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] De la part de Mark Jose Envoyé : mardi 19 décembre 2006 00:17 À : samba-vms@lists.samba.org Objet : RE: samba share on VMS bound volume set Hi Tim, To quote the VMS wizard: http://h71000.www7.hp.com/wizard/wiz_9470.html The OpenVMS Wizard generally discourages the current and continued use of bound-volume sets, and would strongly discourage new uses. So, perhaps you are only asking for trouble with Samba and bound disks? Regards, Mark -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tim Oakley Sent: Tuesday, 19 December 2006 2:19 AM To: samba-vms@lists.samba.org Subject: samba share on VMS bound volume set Hi, SAMBA/VMS version 2.2.8 running on OVMS 7.2-2 has anyone got any experience of using samba shares on VMS bound volumes ? i have configured a share on a VMS bound volume set (2 volumes in the set). Samba works OK but from the windows client side the 'refresh' reacts very strangely. if i create a new file i have to disconnect/reconnect the share in order fro the new file to appear. If i then delete this file it remains visible in the windows explorer until i disconnect/reconnect again. this behaviour is only seen on the volume which is a bound volume server side. any ideas ? thanks. -- # Tim OAKLEY MAURY-IMPRIMEUR SA, Z.I, ROUTE D'ETAMPES 45330 MALESHERBES France Voice: (+33) 02.38.32.34.38 Fax: (+33) 02.38.32.37.72 [EMAIL PROTECTED] www.maury-imprimeur.fr PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html Confidentiality and Privacy Statement This message is intended only for the use of the individual or entity named above and may contain information that is confidential and privileged. If you are not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by telephone and destroy the original message. Box Hill Institute of TAFE is committed to protecting your privacy and the confidentiality and security of personal information provided by you to us. For further information visit www.bhtafe.edu.au or email [EMAIL PROTECTED] PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : honouring default_protection ace
Hi. I must admit that I don't clearly understand myself what is the use of some parameters in the SMB.CONF file. However, after having checked the code, I can say the following: 1. create mask and create mode are synonyms, so if you put both parameters in the SMB.CONF file, only the last one will be effective. 2. The algorithm for setting protection to a created file is : - if inherit permissions is set to true, then the file has the same protection as the directory in which it is created - if false, the protection of the file is set to an OR of create mask/create mode and of force create mode. Don't ask me why. I guess that you may set the force create mode parameter at the global level, to give a minimum access for all the shares, and choose different create mode at the share level to give additional accesses on specific areas. 3. Anyway, because the file protection is fully determined as above, any DEFAULT_PROTECTION ace on the directory has no effect. 4. You can add aces like IDENTIFIER=...,OPTIONS=DEFAULT,ACCESS=... on the directory, and it will be effective. 5. I don't know what SAMBA_ALTERNATE_DIRECTORY_PROTECTION is. I don't see any reference to that anywhere in the code. I hope this will make it a little more clear. JYC -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] De la part de Tom Garcia Envoyé : jeudi 5 janvier 2006 21:16 À : samba-vms@lists.samba.org Objet : honouring default_protection ace I am using Samba 2.2.8 from JYC, February 2005 build on VAX/VMS 7.3. Adding a DEFAULT_PROTECTION ace for a directory, regardless of the setting of SAMBA_ALTERNATE_DIRECTORY_PROTECTION, does not appear to have any effect on new files created in that directory via samba. create mask is 0777, as is create mode for the share (I have also tried omitting the create mode entirely). Rather than the OR behaviour described in the documentation for the unix version, it appears to simply apply the create mode precisely. Is this by design, or should I have configured something? Thanks, -- Tom Garcia | [EMAIL PROTECTED] PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Error: failed to change delete on close flag in 2.2.8 (JYC 17-Aug-05)
I remember that those failed to change delete on close flag errors where linked to a more general problem of file interlocks and sharing, which was fixed with the 20050201 version. There was no changes in this area since. I suggest to check that the .TDB files situated in the SAMBA_ROOT:[VAR.LOCKS] area where reinitialized since that version (i.e. their creation dates are later than the installation of the 20050201 version). If they are not, you just have to stop Samba, delete all those [VAR.LOCKS]*.TDB files and restart Samba. Do NOT delete the SECRETS.TDB file situated in the SAMBA_ROOT:[PRIVATE] area. If the problem happens again, please tell me, and I'll make additional research. JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : Samba 2.2.8 (JYC 31-Mar-05 version) problem with record s longer than 1022 bytes
Unfortunately, I have no more VAX available for compiling, and I have no I64 yet. JYC -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] De la part de David J Dachtera Envoyé : jeudi 18 août 2005 02:38 À : [EMAIL PROTECTED] Cc : 'samba-vms@lists.samba.org' Objet : RE : Samba 2.2.8 (JYC 31-Mar-05 version) problem with records longer than 1022 bytes I'm downloading the new .ZIPs as I write this. They used to come with both VAX and Alpha .OLBs (any hope for I64?). Will check into it as soon as the downloads finish... David J Dachtera At 06:58 PM 08/17/2005 -0400, you wrote: At 01:53 PM 8/17/2005 +0200, COLLOT Jean-Yves wrote: A new version (20050817) is available at http://www.pi-net.dyndns.org/anonymous/jyc/ JYC -Message d'origine- De : Rodney Kimber [mailto:[EMAIL PROTECTED] Envoyé : mercredi 17 août 2005 01:57 À : COLLOT Jean-Yves; samba-vms@lists.samba.org Objet : RE: RE : Samba 2.2.8 (JYC 31-Mar-05 version) problem with records longer than 1022 bytes Thanks for your prompt response. Unfortunately I don't have the facility to compile. :-(I will have to wait for your next release. Keep up the great work. Regards. I'm trying to compile 20050531 version with the change you suggested for long records on my VAX (VMS 7.3/DEC C V6.4/TCPware V5.6-2) as I write this. :-( looks like I'll have to start over... So far, I've gotten warnings on 3 modules, about mismatched pointer types. Compiling CLIPRINT in LIBSMB rparam, rprcnt,/* return params, length */ .^ %CC-W-PTRMISMATCH1, In this statement, the referenced type of the pointer value rprcnt is int, which is not compatible with unsigned int because they differ by signed/unsigned attribute. At line number 91 in DU0:[SAMBA.SOURCE.LIBSMB]CLIPRINT.C;4. is a typical example. The variables are rdrcnt and rprcnt in CLIPRINT and CLIRAP, and rparam_count and rdata_count in CLISECDESC. Since these are warnings, I don't think it will effect the link, though I have seen cases of C compiler warnings that produced unlinkable object modules (or maybe there's a qualifier on LINK that says it's okay to have warnings, and whatever I was building didn't use the qualifier? I don't see such a qualifier in HELP LINK, but maybe there is something in the options file?) I just recompiled the same sources on an Alpha (V7.3-2, DEC C V6.5/TCPware V5.6-2) and had no warnings. DEC C 6.4 vs. 6.5 or an Alpha/VAX difference? If you need, I would be happy to upload the VAX object modules/libraries, assuming it builds/links okay. Or would it be better to do over with the latest sources? John Santos PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : Samba 2.2.8 (JYC 31-Mar-05 version) problem with record s longer than 1022 bytes
A new version (20050817) is available at http://www.pi-net.dyndns.org/anonymous/jyc/ JYC -Message d'origine- De : Rodney Kimber [mailto:[EMAIL PROTECTED] Envoyé : mercredi 17 août 2005 01:57 À : COLLOT Jean-Yves; samba-vms@lists.samba.org Objet : RE: RE : Samba 2.2.8 (JYC 31-Mar-05 version) problem with records longer than 1022 bytes Thanks for your prompt response. Unfortunately I don't have the facility to compile. :-(I will have to wait for your next release. Keep up the great work. Regards. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Disk share size reporting error?
Hi. I don't have this problem here (or I don't do what must be done to see it). What would be the right values for total size/free space on your system ? How do you get the wrong values (Explorer/Properties or by other means) ? JY -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] De la part de Bradford J. Hamilton Envoyé : lundi 11 avril 2005 14:53 À : samba-vms@lists.samba.org Objet : Disk share size reporting error? Hello, I've searched the archives, and could not find a reference to my problem. Apologies if this has been answered before. SAMBA 2.2.8 on VMS The total size for my VMS drives, as seen from a Windows (@K and XP) or FreeBSD client, is listed as 20.0Mb, with free space listed as 0 bytes. I have changed the max disk size and block size parameters to 36000 and 512, respectively, and stopped/restarted NMBD and SAMBA, but the only change that I see is that the free space is listed as 512 bytes, instead of 0. Am I missing a parameter somewhere? TIA PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Vim detects a file on a VMS Samba share has changed when it has not
It is true that the VMS modification is changed when the file is opened in read/write mode, even if the file is not actually written. In fact, the modification date is changed when the file is closed. In order to fix that, in version 2.2.8, Samba keeps track if the file is really changed (i.e. written) or not. When the close request is received, and if the file was not really written, then the modification date is forced to the old modification date (the one the file had when opened). Note that the number of revisions field is not reset, so we have a file with a higher number of revisions, but the same modification date. I guess it would be better if the number of revisions was reset too. I'll make that change when I have time. Now, what happens with VIM? For the moment, I really don't know, because I can't reproduce the problem, probably because I don't have the same environment. What I can see when I try to use vim/cream on my systems is: 1. Just after the beginning of the edit, the number of revisions in increased by 2 with vim alone, and by 3 with cream over vim. 2. The number of revisions does not change after that 3. The modification date is correctly reset (i.e. it does not change) whatever the file owner is. If I compare with what happens in Ben Armstrong's environment, I can see the following differences: 1. The number of revisions is increased by 4 as soon as the editing begins 2. The number of revisions seems to be increased continuously 3. When the editor is not the owner of the file, the modification date is not correctly reset. I think that differences 1 and 2 are caused by the fact we don't have the exact same version of vim, and/or we do not have the exact same cream configuration. I suggest to Ben to try with vim alone (i.e. without the cream layer) to see what happens (check that the behaviour is the same as mine). I don't understand difference 3. It could be some privilege or file protection issue. However, I try with users without any privileges, and my file protection, as far as I know, are the same as Ben's, and I can't get the same results. Could you give me all the details about the file protections and the characteristics of your users? If we can't find the reason, I plan to give to Ben a modified module that will do more logging about the modification time reset processing. JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : naming like xxx.5_htm
The current VMS/Swat version and distribution kit changes xxx.n.html to xxx.n_html. In the next distribution, I will change that in order to replace xxx.n.html by xxx_n.html. I know that this is not enough, and that the html files must be modified too, but I hope that this change will be a beginning. JY Collot -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] De la part de John E. Malmberg Envoyé : jeudi 3 février 2005 18:52 À : samba-vms@lists.samba.org Objet : Re: naming like xxx.5_htm gérard calliet wrote: Hello, Where can I find documentation about the choice of naming htm files from the documentation set in the form xxx.n_htm ? The archives of this list, back on September 9, 2004. The choice was made by the DETAR or UNZIP tool when it encounters a filename that can not be represented on an ODS-2 file. And how configure a browser to follow an associate link in a form xxx.n.htm ? It would likely require rebuilding the browser to handle ODS-5 file specifications, including directories. MOZILLA does not currently handle them. https://bugzilla.samba.org/show_bug.cgi?id=1763 The VMS port of SWAT has special case code to handle the current naming, and that would have to be changed if the names were made to comply with ISO-9660 or with what the web browsers on OpenVMS can deal with. ISO-9660 requires an 8.3 naming convention unless the host operating system can deal with extensions to the standard. I would recommend someone putting together a script/program to fix the filenames and links to an iso-9660 compliant format, and change SWAT to also deal with that. The ideal would be if the producer of the SAMBA documentation would make the output comply with ISO-9660. Alternatively all the tools to convert the raw SAMBA documentation to HTML and other formats seem to still be available on OpenVMS in one form or another, it is just a learning curve to package and use them. If anyone wants to volunteer... -John [EMAIL PROTECTED] Personal Opinion Only PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
New release : 20050201
There is a new release available of Samba/VMS 2.2.8 at http://www.pi-net.dyndns.org/anonymous/jyc/ Please read that page to know about the changes/fixes of that new release. JY Collot PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : SAMBA problem report - file deletion failure
Hi. If the problem happens again, it would be most interesting to know if the files were actually NOT deleted, or if they were, but they seem to stay anyway because their names remained in the cache. The best way to know that, obviously, is to check on the VMS side if the file is there or not. JY -Message d'origine- De : RR - Rod Regier [mailto:[EMAIL PROTECTED] Envoyé : mardi 14 décembre 2004 20:21 À : [EMAIL PROTECTED] Cc : RR - Rod Regier Objet : SAMBA problem report - file deletion failure I have encountered a problem that was reproducible on 3 different workstations accessing a common Samba/VMS server. Sample files created using both DOS window and GUI, and deleted using both DOS window and GUI. Confirming message and file display removal occurred normally with the GUI. Subsequent DIR or GUI view refresh showed the file had in fact *not* been deleted. [SAMBA] LOG.workstation, LOG.SMBD and LOG.NMBD were reviewed without any apparent correlation in time between the delete problems and any logged error conditions. SAMBA server subsystem was recycled, problem retested, had disappeared. Problem not reproducible. Alpha/Samba V2.2.8 release 20041021 plus subsequent patches. OpenVMS/Alpha V7.3-2 w/patches, TCPIP 5.4 ECO 2. This report is supplied to primary make users and developers aware of an anomaly, in the hope that others may contribute related reports that would permit subsequent troubleshooting. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Samba/VMS V2.2.8 build # 20041021 bug report
Actually, what Rod said is not exactly true. It depends if the VMS disk is ODS-2 or ODS-5. Both cases give bad results, but not the same ones. On ODS-2, both NEXT. and NEXT.DIR are seen as directories. We see two directories with the same name, Next. The result is that it's impossible to do anything with the next. file, but the contents of the [.NEXT] directory is available. On ODS-5, it's the contrary: we see two NEXT files, and it's impossible to access the contents of the [.NEXT] directory. Rod is probably using ODS-5 disks. Anyway, I'll try to fix that when I have enough time to do so. Rod's idea is fine, but I hope it may be possible to show correctly both the directory and the file. JY -Message d'origine- De : RR - Rod Regier [mailto:[EMAIL PROTECTED] Envoyé : mercredi 24 novembre 2004 22:50 À : [EMAIL PROTECTED] Objet : Samba/VMS V2.2.8 build # 20041021 bug report Presence of an extensionless-file on an OpenVMS directory will block access to a same-named directory tree residing on the same directory. For example: directory [TOP] contains the files: NEXT. NEXT.DIR;1 (a valid directory file) SMB client access to the Samba server on that node to reference [TOP] will not display the presence of the directory [.NEXT] unless the file NEXT. is deleted or renamed. It seems that this behaviour is an artifact of mapping VMS .dir files to the non-VMS extensionless convention for directory names used on Windows and UNIX, causing a namespace clash with actual extensionless files. In OpenVMS, only files with an extension .DIR, a version number of ;1 and the directory file attribute are valid directories. My recommendation would be to have the Samba/VMS implementation treat extensionless files specially in a lookup. If an extensionless file is found, continue looking for a directory of the same name. If a directory is found, show the directory. Thus, the (extensionless) file becomes hidden instead of the directory, which is probably the lesser of two evils, given how uncommon extensionless files tend to be on VMS. -- Rod Regier, Software Development bus: (902)422-1973 x108 Dymaxion Research Ltd., 5515 Cogswell St., fax: (902)421-1267 Halifax, Nova Scotia, B3J 1R2 Canadamailto:[EMAIL PROTECTED] http://www.dymaxion.ca VMS is today what Microsoft wants Windows NT V8.0 to be! Compaq, 22-Sep-1998 PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
Re: SAMBA share access problem report (Dymaxion 17=17275)
Now that I know that the problem appears only when the McAfee antivirus is there, I could fix it. I am sending a corrected version of the OBJ files directly to Rod, a DIFF of the sources directly to John, and I'll include the fix in the next release. If some of you need to have the fix sooner, please ask me directly. JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Infinite recursion in debug.c when debug_level = 5
You are perfectly right about this DEBUG bug. However, in the 2.2.8 code, sys_fstat actually calls vms_fstat. It is vms_fstat which is redirecting that call to vms_stat, causing the questions you ask. I am the guilty one for this. Actually, I had a couple of reasons for doing this, even if they may be bad ones. First of all, one of the major problems I encountered was to find a way for getting accurate file sizes for variable-length or VFC files. Calling standard CRTL stat() or fstat() does not work, because it gives you the actual number of bytes of the file, which is bigger than the real size when dynamically converted to stream (which is done by stm_read). This generated unwanted zeroes at the end of the text when using notepad, wordpad, etc... with variable-length files. I could not find a better way of getting this correct file size that redirecting vms_fstat() to vms_stat(). In addition, redirecting vms_fstat to vms_stat gave me an easy way of taking advantage of the stat() memory cache. Even if we find a way of resolving the file size problem with fstat(), I must say that I doubt that the changes you propose will greatly improve the speed of Samba/Vms : I measured that 1000 calls to getname() takes 0.74 seconds on my Personal Workstation, and 0.23 seconds on a DS20. The file translation is not performed anyway in this case, and the stat cache existence usually greatly improves the speed. But I may be wrong, and I am interested in your results when the changes are done. JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Samba process NMBD halts the system
Yes, I have encountered the problem, and fixed it. The fix will be available in the next release, which will be posted soon. I am waiting for a final confirmation that another problem is fixed too. JY -Message d'origine- De : Petra Wiegmink [mailto:[EMAIL PROTECTED] Envoyé : jeudi 21 octobre 2004 11:13 À : [EMAIL PROTECTED] Objet : Samba process NMBD halts the system Hello, we have installed SAMBA 2.2.8 on an OpenVMS V7.3-1 system. It works alright, but within the last 10 days it happend twice that the NMBD process was using between 55 and 90% of the CPU time continuously which seriously disturbed the other processes on the system. We have stopped NMBD and started it again. It then worked alright again. Has anybody ever had a similar problem? Any suggestions what I should do to prevent that in the future (other than stopping NMBD altogether). Thank you, Petra Wiegmink [EMAIL PROTECTED] or [EMAIL PROTECTED] ROSIN GmbH http://www.rosin.com Vor dem Leetor 26 53545 Linz am Rhein Germany Tel +49 -2644 - 97003-0 Fax +49 -2644 - 97003-32 Mobile +34 - 627 454 458 PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Samba/VMS Version 2.2.8 Build 20041021
Is it easy to find the changed files ? Yes, the changed files are the ones changed since the 8th of September. Is there any point to building on VMS 7.3-1 or later with out large file support ? I guess not, but may be some people should want it (I can't really imagine why). Actually, I must admit that I did not spend too much time thinking about this. I am completely aware that there should be a lot of things to do in order to enhance the compile/link/install procedures, but I had no time enough to do it. JY -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Envoyé : jeudi 21 octobre 2004 18:13 À : [EMAIL PROTECTED] Objet : Re: Samba/VMS Version 2.2.8 Build 20041021 In article [EMAIL PROTECTED], COLLOT Jean-Yves [EMAIL PROTECTED] writes: I have just posted a new release that fixes some problems that happened to some of you. More information and download available, as usual, at http://www.pi-net.dyndns.org/anonymous/jyc/ Is it easy to find the changed files so that I can merge them into my 2.2.12 build? Like using the date to extract them? Also, is there any point to building on VMS 7.3-1 or later with out large file support? If not, then the DCL procedures can detect the host VMS version and make the correct changes. Otherwise a parameter such as P3 could be used to override the default. The objects proceduced by compiling on VMS 7.3-1 can not be relied on to be correct for linking on older versions of VMS unless a very complex development environment is set up, so that should not be a reason for keeping the command procedures separate. -John [EMAIL PROTECTED] Personal Opinion Only PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : samba 2.2.8/time.c Question on VMS specific modification.
Actually, the reason is not a performance one, but the optimized Samba code does not correctly give the tm_isdst and the tm_gmtoff members of the tm struct. The localtime call is not quite expensive on VMS, so I preferred to use the standard DECC version, which is correctly working. JY -Message d'origine- De : John E. Malmberg [mailto:[EMAIL PROTECTED] Envoyé : vendredi 8 octobre 2004 05:27 À : [EMAIL PROTECTED] Objet : samba 2.2.8/time.c Question on VMS specific modification. The module [.lib]time.c has a VMS specific edit to override the SAMBA code that attempts to optimize the localtime() call. Is there a performance reason for this change? Are specific VMS versions affected by this? -John [EMAIL PROTECTED] Personal Opinion Only PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : samba 2.2.8/util_str.c - Why is '$' exempted from '_' replac ement?
Well, basically the $ character has a special signification on Unix which it has not in VMS. Replacing $ with _ may be useful in a number of cases, but just let's take an example where it must not be done. If you keep the default parameters for a print queue, the print command should be print %f/queue=%p/delete/passall. So, now, try to create a printer share named SYS$PRINT, or try, on any printer share, to print a file named TOTO$TEXT.LOG. If you remove the '$' exemption in UTIL_SRC.C, the %f part of the command will be replaced by TOTO_TEXT.LOG, and the queue name (%p) will be replaced by SYS_PRINT. It will not work so well... JY -Message d'origine- De : John E. Malmberg [mailto:[EMAIL PROTECTED] Envoyé : vendredi 8 octobre 2004 06:21 À : [EMAIL PROTECTED] Objet : samba 2.2.8/util_str.c - Why is '$' exempted from '_' replacement? The module [.lib]time.c has a VMS specific edit to exempt the '$' character from being replaced with an underscore. So far I can not find a reason that VMS needs this change. -John [EMAIL PROTECTED] Personal Opinion Only PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Samba 2.2.8/smbd/server.c - Setting logfile directory from C LI disabled.
For this one, there is no reason I can remember. In addition of not having any reason, this is clearly a bug. If you use the -l option, you just don't get any log.smbd file !!! I guess (and hope) that this option is rarely used. Sorry about that. I'll remove that specific change. JY -Message d'origine- De : John E. Malmberg [mailto:[EMAIL PROTECTED] Envoyé : lundi 11 octobre 2004 05:13 À : [EMAIL PROTECTED] Objet : Samba 2.2.8/smbd/server.c - Setting logfile directory from CLI disabled. The module [.smbd]server.c has a VMS specific change to prevent changing the logfile directory from the SMBD command. Is there any reason that this is done? -John [EMAIL PROTECTED] Personal Opinion Only PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
Re: samba 2.2.8/time.c Question on VMS specific modification.
Can you set your mailer to not put a space the RE tag before the :? It should be the vendor's default setting. I understand the problem, but I really don't know how to do that. My mailer is Outlook 2002 SP2, and I wandered through most of the menus without finding where I can change it. If anyone knows, please tell me. In the meantime, I'll remove manually the blank between RE and :. Please don't be too angry if I forget sometimes. Now I need to find out if this is a VMS C RTL bug or a SAMBA bug, as there should not be a difference. It's definitely a SAMBA bug. The idea is quite simple: the first time you call the optimized localtime(), it construct a table of the different offsets, in seconds, between the localtime and the gmttime for now and some time in the future. It does that by calling the standard C library localtime(), by the way. So, for subsequent calls, it checks the table, gets the current offset, then calls gmtime() after having added that offset to the gmt time. Of course, with gmtime(), the tm_isdst and the tm_gmtoff members of the tm struct are always null. Setting the correct tm_gmtoff is easy: it's the offset from the table. Setting correctly the tm_isdst is probably more difficult. But is this member used anywhere in Samba? Actually, I had a second reason not to use that optimized thing. In the original code, it is implemented by Localtime replacing localtime. With the VMS C compiler, you can deal with such things (using /NAME=AS_IS), but in my opinion, it is a real pain, and you may easily introduce a number of fine bugs by using such a feature in VMS in such a complicated software, especially when you use that with a standard C RTL name. So, I had a choice : make 2 or 3 specific VMS changes in order to replace Localtime by something like samba_localtime, or only one to completely disable that optimized feature, that is a little buggy anyway, and does not actually optimize anything in VMS (I experienced that 5 calls to standard localtime() take 0.50 sec CPU). You know what my choice was. JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : WinXP - OpenVMS tests reproduced using C++ test pro
The stat() or fstat() functions should return the correct results of the real size of the file. In the standard fields, they should have the highest byte written in the file. I am not sure it should. Anyway, it does not and, as far as I know, it never did in DECC or VAXC RTL history. If the stat()/fstat() calls are not doing this, then a small reproducer needs to be submitted to HP so that the stat() call can be fixed. Here it is: -CUT HERE-- #include stdlib.h #include stdio.h #include stdio.h #include stat.h main() { char buffer[1024]; FILE *tst; int i; int n; struct stat st; tst = fopen (TST.DAT, w); for (i=0; i10; i++) { n = fwrite (buffer, 1024, 1, tst); } n = stat (TST.DAT, st); printf (File Size before close = %d\n,st.st_size); fclose (tst); n = stat (TST.DAT, st); printf (File Size after close = %d\n,st.st_size); exit(0); } -CUT HERE-- My guess is that it should only induce a delay, but likely not as bad as what it is fixing. I agree. JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
A note about HAVE_MMAP
Since Dave Jones provided a new, VMS-specific TDB set of routines, defining HAVE_MMAP or not has no more impact on Samba/VMS. JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : dots in directory names of shares on ODS-5 disks
This will be fixed in the next release. Note that dots in a directory name will only be available on ODS-5 disks. Jean-Yves Collot -Message d'origine- De : Jeffrey Coffield [mailto:[EMAIL PROTECTED] Envoyé : samedi 25 septembre 2004 13:56 À : 'Samba VMS' Objet : dots in directory names of shares on ODS-5 disks I have a customer who wants to put dots in a directory name. Since the share is on an ODS-5 disk, this is possible for VMS but Samba 2.2.8 does not seem to follow the VMS convention of using a ^ in front of a dot. Has anyone looked in to this? If not I may attempt to patch the source code. TIA Jeff Coffield PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : WinXP - OpenVMS tests reproduced using C++ test progra m
Hi. After some research, I have answers about that issue. Actually, when analysing the network dialogues between the Samba/VMS server and the clients, there are two differences: 1. The block size (1024 bytes when XP client, 61 Kbytes when Linux client) 2. The 1-byte writes and QFILE_INFOs requests, (only some with Linux client, very numerous with XP client). I tried to see what happens when using an XP client with a Linux Samba server. The block size was 1024 bytes too, but there was only a couple of 1-byte writes. So, it seemed that the block size was client-dependent, and the numerous 1-byte writes was server-dependent. In order to check that, I made some changes in Ben's C++ bench program. First, I changed the code in order to use binary stream files (using fout.open(dataFile,ios::binary);. With only that change, the block size went up to 4096 Kb. Then, I made additional changes, in order to force a 60 Kbytes buffer for the stream file (using setvbuf()). As I expected, the block size was now 60 Kbytes. My conclusion is that the block size is only due to the client. In the original bench code, the stream file buffer is left to the system default. It looks like the system default is 61 Kb on Linux, and either 1Kb or 4Kb on XP, depending upon the text/binary characteristic of the stream. This point is by far the main reason of the poor performances. The bench elapsed time goes from 45 seconds down to 6 seconds by adding the 60 Kbytes buffer. Unfortunately, I don't know if it is possible to change the XP default buffer size at system level. I made some research, but I could not find anything. Now let's see the 1-byte writes and QFILE_INFOs requests. It looks like those actions are done by the client in order to provoke extensions of the allocated file space, before actually writing into it. Unfortunately, on VMS, the real size of the file is not visible until the file is closed. That can be seen by typing DIR/SIZE=ALL while the file is being written. It gives 0/. So, when the client asks for an extension (Write 1 byte) then checks the new file size (QFILE_INFO), it always gets 0 for answer. That is quite perturbing for him, so it goes on trying to extend the file. I made a small change on the VMS server side: when the write request is only 1 byte long, I close and reopen the file. This updates the external view of the file size. With that change, the numerous 1-byte writes and QFILE_INFOs disappear, and the dialog is exactly the same with XP client that it is with the Linux client. The time is the same too: 3.5 seconds. I'll make some additional testing to check that this Close/Reopen does not have side effects, and if it does not, I'll include that change in the next release. Jean-Yves PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Packet analysis: WinXP vs. Linux-VMS shows dramatic differ ences
Hi. I was very interested (and very surprised) by the experiments and results described in the last messages, so I tried to reproduce the issue myself. Note that, instead of using Ethereal or other utilities, you can run smbd with log level 3, and all the transactions are logged. I downloaded the ruby stuff, and I ran the bench provided by Ben Armstrong. I saw exactly the behaviour described by Ben (1024 bytes writes, 1 byte writes, QUERY_FILE_INFOs, etc..). The result is incredibly slow. I did some other tests with a simple copy DOS command, and with copy/paste clicks in the Windows Explorer. Those times, there was only 61 Kb writes. No more 1024 bytes writes, no more 1 byte writes, no more QUERY_FILE_INFOs, and good performances. I then used a utility (http://www.pc-tools.net/win32/ptime/) that you may know, and that enables you to get the execution time of a command. Here are the results, for a COPY BENCH.TMP Z:\BENCH.TMP command : - BENCH.TMP is a 1 Kb local file (the file created by Ben's ruby bench) - My whole network is 100Mbits/Full Duplex - My client is a Windows XP/SP2 PC - when Z: is a Samba/Linux box (bi-processors Intel 2Ghz), the elapsed time is 2.75 seconds - when Z: is a Windows NT server (a quite old one, I don't know exactly the type), the elapsed time is 3.79 seconds - when Z: is a Samba/VMS Alpha DS20-666, the elapsed time is 2.35 seconds As you see, the performances of Samba/VMS are not so bad. My feeling is that there are some strange things happening between Windows/Ruby and the Samba/VMS server. Fixing that will probably be quite difficult, and will probably demand analysis of the Ruby code as much as the Samba code, in order to understand what happens. Is this Ruby thing used by many people? Do any of you have similar very bad performances when writing a file, but with other software? JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Packet analysis: WinXP vs. Linux-VMS shows dramatic differ ences
Once again, I am sorry, but it looks like I don't understand. As far as I know, Ben pointed out a problem of performances when writing files on a Samba/VMS 2.2.8 server from an XP client. I could reproduce that problem by using ruby, but I could not by using anything else. Another way to say is: according to me, the WinXP vs. Linux-VMS shows dramatic differences topic is true only when using ruby. Could someone tell me if I am right or wrong here? When you refer to a Windows NT 4.0 client on SAMBA 2.0.6, or to issues dealing with modify in place stream-CRLF or stream-LF files, you are perfectly right, but I don't clearly see the connection with Ben's problem. Porting Samba 3 or 4 to VMS is obviously a very nice thing to do, but I fear that it will take a lot of work and quite some time. In the meantime, I may be able to understand or even fix Ben's problem on 2.2.8, so I prefer to focus on that specific problem. I have a last question: what is exactly the notepad problem on VMS ? JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : RE : smbd serves connects only when ran in interactive mode (-i)
The point is that, according to the logs, there is more than 6 seconds between the first line of the log (Samba starts) and the error message ( socket not connected). I could reproduce the same socket not connected error by adding a 5 sec sleep in the Samba code. Services such as FTP do not have much to do for initializing, so they answer to the initial connect request rapidly. I guess there should be some registry key somewhere on the PC that could be changed in order to have a longer time-out, but I don't know where. Anyway, it's a good idea to try TCP/IP 5.3, but I don't have too much hope. JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Scripting task to improve reading of documentation on VMS
This would indicate that something in SWAT is doing a translation. Yes, sure, something does. A piece of VMS-specific code in CGI.C keeps only the first dot(.) and replaces all subsequent dots by underscores(_). After that, I renamed the files (i.e. SMBD.8.HTML to SMBD.8_HTML). JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : smbd serves connects only when ran in interactive mode (-i)
Well, this is quite surprising. Could you give me some more information ? 1. The results of $ TCPIP SHOW SERVICE/FULL SMBD 2. A copy of your SMB.CONF file 3. Set the log level to 3, delete SAMBA_ROOT:[VAR]LOG.*;*, retry to access a share from your PC, and please send me the full files LOG.SMBD and LOG.your-pc (not only excerpts, please). Thanks. JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : smbd serves connects only when ran in interactive mode (-i)
A couple of other questions : - What is the BYTLM quota of the user associated with the TCPIP SMBD Service ? - what is the value of the SYSGEN parameters PQL_MBYTLM ? - Can you correctly use SWAT from the PC or not ? (i.e. http://your-server:901) JY -Message d'origine- De : Ashot Bord [mailto:[EMAIL PROTECTED] Envoyé : lundi 13 septembre 2004 19:39 À : [EMAIL PROTECTED] Objet : Re:RE : smbd serves connects only when ran in interactive mode (-i) Here's the excerpts from log-files ( I set the SMBD debug level to 7), the smb.conf debug level I set to 0, to rule out additional disk IO related lags to write log-recs. == log.smbd [2004/09/11 23:45:51, 0] DISK$DATA:[SAMBA-2_2_8-SRC.SOURCE.LIB]UTIL_SOCK.C;5:(1¦ getpeername failed. Error was socket is not connected == smbd_startup.log -- Adding chars 0x9c 0x0 (l-u = False) (u-l = False) load_unicode_map: loading unicode map for codepage 850. load_unicode_map: loading unicode map for codepage ISO8859-1. loaded services Open file /samba_root/var/locks - /samba_root/var/locks Open file /samba_root/var/locks in record mode stm_read: request to read 62000 bytes VMS file size /samba_root/var/locks, rfm = 2, size = 0 claiming 0 getpeername failed. Error was socket is not connected Open file /samba_root/private/HOME.VAXNODE.mac - /samba_root/private/HOME__2EVAXNODE.mac === notice the claiming 0, followed by getpeername failed msg -- that's where the difference is -- in interative mode, it gets the socket (no getpeername failed) and goes on with establishing the connection, in service mode it's about the end. what's so different in interactive mode about socket negotiation, than it is in service mode? once again, the conf is simple and the connection works when mapping from VAX to VAX, even in service mode. The problem is with WIN to VAX in service mode. AB PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : smbd serves connects only when ran in interactive mode (-i)
Is there anything in the LOG.SMBD and/or LOG.your-pc files in the SAMBA_ROOT:[VAR] area ? What is the log level in SMB.CONF ? JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : New Samba/VMS Release : build 20040908
Hi. 1. I can not get guest access to work. I think it works here, but may be I am not doing what you want to do. What I do is : - define SAMBA__GUEST as the guest account in SMB.CONF - define Bad user for the map to guest parameter (Never is the default) - define guest ok as Yes in some shares, and as No (the default) in others And it works. In addition, if I define 2 as log level, I have no timeout problem, and I get messages in the log such as : connect to service share1 as user samba__guest (for guest ok = Yes shares) Invalid username/password for share2 [samba__guest] (for guest ok = No shares) I hope that it's decision is based on a file extension (about the VMS Record Format parameter) Actually, I added that feature because Mr Tapani Rundgren ([EMAIL PROTECTED]) asked for it. Mr Rundgren is using a software product that writes files on Samba/VMS shares, but separates the written lines by LF instead of CRLF. This is the case, for example, if you use smbmount from a Linux box, and create text files with vi or any Linux tool. When a VMS program or procedure tries to read such files, it does not see any line terminator, and sees the files as if they have only one big line. The parameter gives the opportunity of telling VMS that the line terminator is a single LF. This is a parameter attached to the entire share (a similar option exists in Pathworks). I don't understand what you mean by basing the feature on a file extension. In my example (Samba share accessed thru Linux smbmount, and accessed thru some Windows box at the same time), a text file will have only LF when created from Linux, but will have CRLF when created from Windows. I think that you can't securely assume that a given file comes from Linux because, for example, its extension is .lis, or from Windows because its extension is .txt JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
New Samba/VMS Release : build 20040908
As you see, with this release I start a new policy of naming the kit files. I hope it will satisfy those of you who asked for an easiest way of managing the Samba/VMS versions. The kits will now be named SAMBA-2_2_8-specific-buildnumber.zip, where : - specific can be : o OBJ: Objects files for all Alpha/VMS versions. o VMS73-OBJ: Objects files for Alpha/VMS version 7.3-1 and above. Supports big files. o OBJ-VAX : Objects files for all VAX/VMS versions. o SRC : Sources and procedures, all Alpha/VMS and VAX/VMS versions. o VMS73-SRC : Sources and procedures, Alpha/VMS version 7.3-1 and above - buildnumber will be formed with the release date (i.e. 20040824 for the 24-AUG-2004). There will never exist more than one kit per day In addition, for a better compatibility, the backups are now created with the /INTERCHANGE option. Now, the news. 1. A new VMS Record Format share parameter has been added. It defines the RMS record type of the files created by Samba, i.e. the way VMS will see those files. The default is stream, but it can be forced to stream_lf. Support of this parameter has been added in SWAT, at the end of the Miscellaneous Options (Advanced View). This feature can be useful in some specific situations, when the records written by the clients are separated with LF, instead of the standard Windows CRLF, and when those files must be processed by VMS programs or procedures. 2. A procedure MAKE_SAMBA_GUEST_ACCOUNT.COM has been added in the [BIN] area. It creates a Samba Guest account named SAMBA__GUEST (taken from Samba/VMS 2.0.6, thanks to John E. Malmberg) 3. A procedure SAMBA_COMMANDS.COM has been added in the [BIN] area. It creates DCL symbols for an easier way of using the Samba utilities. (thanks to John E. Malmberg too) 4. The template SMB.CONF has been reviewed to be compatible with version 2.2.8. It gives an example of a VMS server within a Windows Domain. 5. The default debug level of some utilities (SMBCLIENT,NMBLOOKUP) has been reset to 0 6. An erroneous message in the sub-feature of sending messages with SMBCLIENT has been corrected. 7. The code has been checked in order to remove the /STANDARD=VAXC compiler option in COMPILE.COM, and to remove irrelevant warnings and informations. However, depending upon the version of the DECC compiler you use, you may encounter a lot of informational messages if you choose to recompile Samba/VMS from the sources. You may prefer to use the old procedure, which is now included under the name COMPILE_STD_VAXC.COM Once again, thanks to those of you that asked for those enhancements, or pointed out the problems that I tried to fix. JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : Link problems with V2.2.8
I posted on the download site (http://www.pi-net.dyndns.org/anonymous/jyc/) the file containing the missing modules, and some explanation about using it, for those of you who use old VMS versions. JYC -Message d'origine- De : Leo Klein [mailto:[EMAIL PROTECTED] Envoyé : samedi 19 juin 2004 09:31 À : [EMAIL PROTECTED] Objet : Re: RE : Link problems with V2.2.8 Thanks to John E. Malmberg for your very helpful posting, which explained what was wrong and what needed to be done. Unfortunately, I've decided not to continue with my efforts at linking Samba: though I understand I need to write my own substitutes for the two missing SNPRINTF routines, I haven't a clue about their parameters, return values and processing required, and anyway I don't have the Dec C compiler - that's why I wanted to link object code rather than building from sources. STOP PRESS: Jean-Yves Collot has just sent me a rebuilt snprintf.obj which solves my undefined symbol problem, and the link now succeeds. John E. Malmberg wrote: Leo Klein wrote: $ @link Linking SMBD %LINK-W-NUDFSYMS, 2 undefined symbols: %LINK-I-UDFSYM, DECC$GXSNPRINTF %LINK-I-UDFSYM, DECC$GXVSNPRINTF Several of the newer xxxSNPRINTF variants are not in the older CRTL and are being added to the newer versions. For older CRTLs, you must supply replacement routines. When supplying the replacement routine for a standard C library, it must not have the same public symbol name as the routines, as this will cause problems. Some of these problems will be visible at link time, some will not, and may take quite a bit of effort to find why the code is malfunctioning. The GXSNPRINTF and GXVSNPRINTF calls can be generated by the compiler for a variety of public C RTL routines depending on your optimization settings, so you have to look at the source modules. It should not be hard to write replacement routines, and likely they are already present in the SAMBA code, and a change to config.h will make them active. For example, a missing VSNPRINTF routine would be replaced with a routine named rep_vsnprintf, or samba_vsnprintf, or my_vsnprintf. In the config.h there would be an option: #define HAVE_VSNPRINTF or #undef HAVE_VSNPRINTF Depending on if your platform supports that call. In one of the header files, or in the modules that use vsnprintf(), there would be the following conditional code, or something similar. #ifndef HAVE_VSNPRINTF #define vsnprinf samba_vsnprinf #endif If you need to supply your own replacement routine, then the above conditional code would be put in the CONFIG.H file to minimize edits to the common UNIX SAMBA code. And even though it seems to be an easy thing to do, do not ever name the replacement routine the same as a standard C library function. A good optimizing C complier knows about many of the library routines and will in line them, so if you are trying to change the behavior of a standard function, the compiler may not realize that, and inline the standard function. Also the link time substitution of user supplied routines covering up system libraries only works reliably on platforms that do not use shared images for their libraries. People who do not heed the above warning usually end up with others that try to build their code posting on comp.os.vms trying to find out why they are getting weird build errors after an OpenVMS upgrade or C RTL ECO. Anything that is using standard C function names for their own public symbols is virtually guaranteed to eventually not build on OpenVMS. And several people have posted on the SAMBA Technical list for various UNIX platforms that have the same problem when SAMBA did the same thing. -John [EMAIL PROTECTED] Personal Opinion Only PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Link problems with V2.2.8
Michael Lemke wrote : Well, for what it's worth, I had the same problem a while ago. I solved it by rebuilding from source and modifying config.h or whatever that was until all the unsupported *printf routines wheren't needed. Yes, this is a good solution, if you have the DECC compiler available. You just have to comment out, or delete, the following lines in [.VMS]CONFIG.H #define HAVE_SNPRINTF #define HAVE_C99_SNPRINTF #define HAVE_VSNPRINTF #define HAVE_C99_VSNPRINTF Then, recompile the SNPRINTF.C module, with the following command : $ @[.VMS]COMPILE NODEB SNPRINTF And run the LINK procedure. JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Link problems with V2.2.8
No, you can't ignore those link messages. Samba won't work. Please download from HP and install the latest available CRTL patch kit for your VMS version, and it should be OK to link and run Samba. JY -Message d'origine- De : Leo Klein [mailto:[EMAIL PROTECTED] Envoyé : mardi 15 juin 2004 12:26 À : [EMAIL PROTECTED] Objet : Link problems with V2.2.8 I'm having a problem linking version 2.2.8: there are many messages about two undefined symbols. I don't suppose I can ignore them and go ahead with the produced .exe files as though all is well. I'm using two files downloaded yesterday from http://www.pi-net.dyndns.org/anonymous/jyc/ and dated 10-May-2004; they are samba-2_2_8-src.zip and samba-2_2_8-obj.zip. I don't have the Dec C compiler. $ @link Linking SMBD %LINK-W-NUDFSYMS, 2 undefined symbols: %LINK-I-UDFSYM, DECC$GXSNPRINTF %LINK-I-UDFSYM, DECC$GXVSNPRINTF %LINK-W-USEUNDEF, undefined symbol DECC$GXSNPRINTF referenced in psect $LINK$ offset %X06A0 in module SERVER file DKB400:[SAMBA228.SOURCE.SMBD]SERVER.OBJ;2 %LINK-W-USEUNDEF, undefined symbol DECC$GXVSNPRINTF referenced in psect $LINK$ offset %X00A0 in module DEBUG file DKB400:[SAMBA228.SOURCE.BIN]SAMBA.OLB;3 and so on. There was a similar problem with a different PRINTF symbol in September 2002, and Jean-Yves fixed it. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
New Samba/VMS version for Big Files support
I posted on the usual http://www.pi-net.dyndns.org/anonymous/jyc/ location a new couple of kits that supports big files (i.e. files bigger than 2 Gigabytes). Except this feature, there is no differences between this kit and the standard one, so if your files are not so big, you have no reason to update. Please note that this big files version works only on Alpha, with VMS 7.3-1 or later. Many thanks to James Ziller, who originally popped up the problem, and was a great help for testing. Jean-Yves Collot PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : 2 GB limit?
I started to have a look on this 2GB size limit, and I don't clearly understand the last couple of messages in this list, about the _LARGEFILE definition. I suppose that you make references to some SAMBA/VMS version older than the one (2.2.8) on which I am working. Having made a couple of tests, and looked around in the source, I think that it would not be a big problem to support files between 2Gbytes and 4Gbytes (i.e. the size in bytes can be coded on an unsigned long), providing that your version of the C RTL is recent. Actually, I don't really know which version is OK. I know that VMS 7.3-1 is OK, but 7.1-2 is not. For files bigger that this 4Gb max, the problem is far more difficult. JYC PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : SMBMOUNT
I don't. In my opinion, it is far too much complicated to achieve, considering the time I can spend on it. JY Collot -Message d'origine- De : Stephen Eickhoff [mailto:[EMAIL PROTECTED] Envoyé : mercredi 10 mars 2004 03:45 À : [EMAIL PROTECTED] Objet : SMBMOUNT Is there anyone currently working on getting SMBMOUNT support in the VMS port? -- Stephen Eickhoff www.operagost.com -- PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Tuning for performance
Actually, as you may guess, the bad performance is due to the number of files in a directory (in fact, the size of the directory (.DIR) file), not of the size of the files themselves. Because of the RMS directory caching feature, the real gap is when the directory file is bigger than 127 blocks. The bad performances in such a situation (huge .DIR file(s)) are not limited to Samba, however. Even a simple DIR DCL command may be quite slow. When you browse a Samba shared directory through the Windows Explorer, it's more or less the same as executing a DIR/SIZE command on this directory. Is Samba much slower than DIR/SIZE ? The only thing I know that can enhance the performance problem in that case is to increase the ACP_DIRCACHE sysgen parameter, in order to make it big enough to cache the whole big directory. The first access to the directory is slow, but subsequent accesses are notably better. Jean-Yves -Message d'origine- De : Profaizer, Joe [mailto:[EMAIL PROTECTED] Envoyé : lundi 2 février 2004 20:28 À : '[EMAIL PROTECTED]' Objet : Tuning for performance I just installed samba 2.2.8 and I have a subdirectory that is very large (100GB of many files). I'm curious as to what performance tuning parmeters I can adjust to help performance. The performance is extremely slow. Thanks, ..Joe PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : swat on VMS
Sure, it works well. The SWAT program is implemented as a TCP/IP service on port 901, so if the SWAT service is enabled, you just have to connect to it on http://your-node:901 JY -Message d'origine- De : Hank Vander Waal [mailto:[EMAIL PROTECTED] Envoyé : mardi 27 janvier 2004 03:02 À : [EMAIL PROTECTED] Objet : swat on VMS anyone get this to run ?? Do I need a web server running on the VMS system ?? TIA PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Curious lock problem
This is an issue already fixed in the last Samba/VMS version -Message d'origine- De : Terrence Branscombe [mailto:[EMAIL PROTECTED] Envoyé : mercredi 12 novembre 2003 03:32 À : [EMAIL PROTECTED] Objet : Curious lock problem PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : New samba not starting
Hi. That's my fault. For performance reasons, I changed the default Samba/VMS option for oplocks to No. I must admit I don't clearly understand what those oplocks are and what they are supposed to do. I just saw that when enabled, they add activity of IP messages between the smbd processes on the VMS server, and the overall performance is worst. I know that the Samba documentation claims that The oplock code can dramatically (approx. 30% or more) improve the speed of access to files on Samba servers. In my experience, it may be true on Unix, but it's not true on VMS. If I'd known better, I would have changed the level2 oplocks default to No too, in order to avoid the warnings in TESTPARM. I'll do it for the next release. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : acls and smb passwd file
The problem here is that, for performance reasons, some kernel locks are used by the SMBD process, and those locks are created when calling stat(). Unfortunately, stat() is called by other Samba components (such as smbpasswd), and it takes (or tries to take) the same kernel locks, even if there is no performance issues. The result is that most, if not all, components of Samba/VMS does not work properly if the user has not the CMKRNL privilege (and probably a couple of other ones, such as SYSLCK). I am going to work on this, in order for the locks to be taken only by the SMBD processes. JY Collot -Message d'origine- De : system manager [mailto:[EMAIL PROTECTED] Envoyé : vendredi 25 juillet 2003 07:34 À : [EMAIL PROTECTED] Objet : acls and smb passwd file Hello, If I try to run samba_root:[bin]smbpasswd as a non-priv.ed user I get : Error Lock Volume F11B$vUSER1 : insufficient privilege or object protection violation repeated ten times and then Old SMB password: New SMB password: Retype new SMB password: machine 127.0.0.1 rejected the session setup. Error was : Call returned zero bytes (EOF) . Failed to change password for MORPHIS where it doesn't matter what I put in for the passwords, except that if the two new passwords don't match it rejects me for that. USER1 is the label of the disk that happens to be my default and of course MORPHIS is my username. This appears to occur in [.SOURCE.VMS]VMS_SUPPORT.C in char *getpass(char *prompt) [big snip] new_cache-label[volnamsize] = 0; strcpy (new_cache-resname,F11B$v); strcat (new_cache-resname,new_cache-label); for (i=0;i18;i++) { if (new_cache-resname[i] == 0) new_cache-resname[i] = ' '; } /* Put this new cache in the list */ if (first_cache == NULL) first_cache = new_cache; else { cur_cache = first_cache; while (cur_cache-next != NULL) cur_cache = cur_cache - next; cur_cache-next = new_cache; } /* Get infos about the RSB of the volume lock */ cur_cache = new_cache; } sts = sys$cmkrnl (update_lock,0); if ((sts 1) != 1) DEBUG(0,( Error Lock Volume %s : %s\n,cur_cache-resname, str_cache-resname, strerror(EVMSERR,sts))); OTOH if I set proc/priv=nocmkrnl I get the same error messages but the password successfully changes. Changing permissions on the directory and teh file with passwords doesn't seem to do any good. Why is it trying to lock a volume? Why is it trying to create cache based on where the user is sitting? - when I run testparm I get: WARNING: lock directory /samba_root/var/locks should have permissions 0755 for browsing to work SAMBA_ROOT:[VAR]LOCKS.DIR;1 (RWE,RWE,RE,RW) I tried changing it to w:re but no change. - Perhaps this is entirely unrelated but if I do $ mcr authorize sho system/all system has the following identifier SAMBA_ROOT %X80010017 which I saw during the install process. When I do $ dir/secu samba_root:[00...]*.dir I see something like the following (much edited) Directory SAMBA_ROOT:[00] BIN.DIR;1[SYSTEM] (RWE,RWE,RE,RE) (DEFAULT_PROTECTION,SYSTEM:RWED,OWNER:RWED,GROUP:RW,WORLD:RE) LIB.DIR;1[SYSTEM] (RWE,RWE,RE,RE) (DEFAULT_PROTECTION,SYSTEM:RWED,OWNER:RWED,GROUP:RW,WORLD:RE) PRIVATE.DIR;1[SYSTEM] (RWE,RWE,RE,RW) (DEFAULT_PROTECTION,SYSTEM:RWED,OWNER:RWED,GROUP:RW,WORLD:RW) SWAT.DIR;1 [SYSTEM] (RWE,RWE,RWE,) (DEFAULT_PROTECTION,SYSTEM:RWED,OWNER:RWED,GROUP:RW,WORLD:RE) TMP.DIR;1[SYSTEM] (RWE,RWE,RE,RW) (DEFAULT_PROTECTION,SYSTEM:RWED,OWNER:RWED,GROUP:RW,WORLD:RW) VAR.DIR;1[SYSTEM] (RWE,RWE,RE,RW) (IDENTIFIER=*,OPTIONS=DEFAULT,ACCESS=READ+WRITE+CONTROL) (DEFAULT_PROTECTION,SYSTEM:RWED,OWNER:RWED,GROUP:RW,WORLD:RW) The files in [var] have the same identifier. The files in [swat] look like: SAM.AA01_GIF;1 [SYSTEM] (RWED,RWED,RE,) (IDENTIFIER=%X80010031,ACCESS=READ+WRITE+EXECUTE+DELETE+CONTROL) If I do UAF sho /id/value=(id:%X10031) %UAF-E-SHOWERR, unable to complete SHOW command -SYSTEM-F-NOSUCHID, unknown rights identifier smb.conf [global] workgroup = PHYSICS dead time = 10 map archive = no printing = bsd printcap name = /samba_root/lib/dummyprintcap.dat load printers = yes print command = print %f/queue=%p/delete/passall/name=%s lprm command = delete/entry=%j security = user smb passwd file = /samba_root/private/smbpasswd.dat encrypt passwords = yes default service =
RE : vms_tdb.c modifications
Your TDB changes will be included in my next version. Thanks. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : Problem with VMS_SUPPORT.C
These open/writes should not update the file modification dates. The only way I see around this is to track to see if the file was actually modified and then use the XQP function to restore the dates to what they were when the file was opened. I completely agree. That's exactly what the last Samba/VMS version does. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE: RE : smbfs reads Variable files incorrectly from VMS Samb
Why shouldn't Samba handle files in the same way that most FTP servers on VMS do: use RMS to read the records and push them to the client? It's basically what Samba does, but, for the moment, only when the variable-record file is opened in Read-only mode, and this single test was OK when dealing with native Windows clients. Remember that FTP has two modes (ASCII and BINARY), and that usually the user has two choose himself which mode is appropriate for the file he wants to upload or download. Samba has to decide itself which mode is the good one. In addition, both cooperating FTP programs are doing the same thing, in ASCII mode : they send and receive records. This is not the case with Samba. When Wordpad reads the file, for example, it must be pushed to it as records separated by CR/LF, but when it writes it back, it sends it to Samba as big buffers, the end of one buffer is not always the end of a record. Anyway, I am going well, and the smbfs issue will be fixed soon. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Possible TDB/Samba optimizations.
Connections is one that I would want to move to use the LOCKING code. Connections must be removed from the table if a SMBD process crashed. Removal of old connections from crashed SMBD processes is now done : when a new connection comes in, all connections in the TDB file that is owned by non existent processes are removed. I think that there is a way to have more bytes than that associated with a lock. It would take some research. Possibly sublocks. As an example, the TDB record for a connection is 604 bytes long. For the LOCKING TDB file, it is 145 bytes long. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Problem with VMS_SUPPORT.C
Hi. Could you give me a copy of the full line (or lines) of the log where the error appears ? It should look like $GETDVI ERROR for xxx: sts = nnn, iosb=yyy You are right that GETDVIW should be better than GETDVI, but this is clearly not the point. In addition, you CAN use a full file spec for $GETDVI, so it's not the point either. The addressing mode of the iosb is OK too. Actually, there is probably something specific on your site that makes some bug appear, because this problem does not appear here or anywhere else as far as I know. Anyway, I need the full error message to work on it. Why would you need to retrieve this information when reading files? It seems to me that SAMBA is doing a lot of unnecessary work here. Well, it is very true that SOMEONE is doing unnecessary work, but it's not SAMBA. Samba is just a server, and does what the client asks it to do. If Samba retrieves the nb of free blocks, the only reason is that the client asked for it. Sure, the client has no understandable reason to do so, but it did. I think that in the Windows World, there is enough wonder in just observing what is done. Trying to understand WHY it's done is far too much for me. As an example, when you just right-click Properties on a file on the XP Explorer, it sends to Samba more than 10 consecutive requests to open that file, and one of those open requests is in WRITE mode ! PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
New VMS 2.2.8 Version
The 2.2.8 version for VMS is available at http://www.pi-net.dyndns.org/anonymous/jyc/ It includes the Unix version 2.2.8, and some specific VMS fixes : - Unexpected changes of file structure (VARIABLE to STREAM) and modification date - Correct display of the file dates (no more differences between VMS and PC clients) - Fix of SMBD crashes when sharing 00 directory of a disk - Automatic creation of TDB files when needed (thanks to Dave Jones) PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : Detecting directory changes.
Doing whatever F11X$POSIX_FASTRDSEQNOS does. This is one of two things - it either enqueues a lock and dequeues the lock (...) , or it directly accesses the memory structures used by the lock manager The routine actually enqueues/dequeues a lock, and does not access directly the lock manager data structures. That's true is not so FAST, but I assume it does this because it does not want too be too much dependant of the internal data structures, which are regularly changing when VMS version changes. I tend to prefer notification via AST over polling. It is more efficient and should be faster I do too, but may be not in this case, because 1. doing this may interfer with internal VMS behaviour, and slow down the entire system (imagine you have a big lot of SMBD processes), 2. you may end with a huge number of locks (one per directory/per SMBD process) 3. using blocking ASTs in kernel mode can be highly dangerous, because image exit does NOT dequeue those locks, so you must be absolutely sure that there is no kernel blocking AST left when you exit, unless you'll crash when trying to execute a blocking AST routine that is no more in memory. (about rewriting the F11X$ routine in C) Since this could work on a VAX, but the routine doesn't exist there, then it might be useful if you want it to work equally well on a VAX. That's true, and would make it a little less undocumented/unsupported, because it would only depend on the resource names used by the XQP, and not on the actual presence of the F11X$ routine. I'll do it as soon as I have a little time. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : stat (AGAIN!)
Using your method instead of opendir() readdir() seems very interesting, and will certainly give better performances. When your version of VMS_OPENDIR works, I'll be happy to put it in the distribution. Please check that it works too with ODS-5 file system, with the extended file names ! For information, I checked the efficiency of the directory/stat memory caching. It's about the same for both caches. Even when doing numerous changes (file creations/deletions) the number of disk I/Os is divided by about 4. When only browsing, it's obviously far more ( 10) . PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Detecting directory changes.
Without knowing what is using the internal routine, I do not know if it will remain, will be removed, or if changes to RMS will leave it totally or partially non-functional. Actually, this internal routine uses a feature of the XQP, not of RMS, but it's a detail. It may work for now, but as it is an internal routine, it is unsupported, and may break with out notice, even from the installation of an ECO kit. Well, this is hardly what happened in the past, however. As Dave Jones pointed out, this internal routine is dated 1993 in the sources, and uses an XQP feature (using the VMS lock manager in a certain way) that seems to be present since the very beginning of the XQP itself (middle of the 80s, if I remember correctly), precisely for managing the validation of the directory XQP cache in a cluster environment. In my opinion, the probability of a sudden break out for this particular feature sounds low. Note that my old method of looking at the lock value block of the XQP volume lock used another XQP feature that is hardly more documented or supported... Caching directory and file (stat) information enhances Samba/VMS performances very much, that's a fact. May be that the price to pay is to take the risk of using some undocumented/unsupported VMS feature. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : PB with filename
We have many filenames with double underscores ( __ ) under VMS and they do not appear under Windows explorer. I have a look on Internet it seem to be normal under Samba Vms !!! The best idea I have is to use ODS-5 file system on your disk. Another way is to define the logical name : $ DEFINE/SYSTEM SAMBA_FILESPEC_ENCODE NEVER If you do that, your files will appear again, but the non-compatible-with-VMS filenames (such as names with blanks inside, or with the ~ character, for example) will not be permitted via Samba/VMS. May be you can accept that. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : RE : Detecting directory changes.
(about F11X$POSIX_FASTRDSEQNOS) : I am trying to find out how stable that internal routine is. I am trying to see if a supported interface can be developed to either give notification on change, or a way to detect changes in the directory. As did Dave Jones (thanks again to him), I looked into the source of this internal VMS routine, and it looks OK to me. As I said in a previous message, it is not supported on VAX (VMS 7.1, which is the only version I have access) : it returns a SYSYEM-E-UNSUPPORTED status. I used Dave's example for testing and I am very satisfied with it, so I now use it very successfully for directory cache validity checking in Samba-Vms (for the VAX version, I keep the 30 seconds validity time-out). PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : passwd()/crypt() emulation incorrect
Dave is right : i twas not correct. I'll correct that for the next release. You'll have to wait a little longer... JY PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : SAMBA 2.2.4 or 2.2.7a and broken pipe
SAMBA 2.0.6 allows the smbd process to be run interactively Sure, 2.2.7 kept that very interesting feature for debugging. The syntax is RUN SAMBA_ROOT:[BIN]SMBD -i -dn n being the debug (trace) level you want. Between 3 and 5 is usually enough to have a lot of information. Before doing that, however, you'll have to $ TCPIP disable service SMBD and to stop any SMBD_* process of the system. Note that this feature is also available for NMBD. PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.catb.org/~esr/faqs/smart-questions.html
RE : Detecting directory changes.
That's just great. Does not seem to work on VAX platforms, however. I'll include that in the new release. Thanks a lot JY Collot -Message d'origine- De : Dave Jones [mailto:[EMAIL PROTECTED] Envoyé : lundi 3 mars 2003 07:57 À : [EMAIL PROTECTED] Objet : Detecting directory changes. I just joined this list, I've been playing with samba to the past few weeks. In the archives there was discussion of the difficulty of reliably invalidating the stat cache when directory updates don't change it's allocation. While poking around the VMS source listings for a better solution, I came across a system routine that looks made to order for this problem. If you open the directory file, the function F11X$POSIX_FASTRDSEQNOS can be used to quickly fetch the header and data sequence numbers stored in the access lock's value block. Any change to the contents of the directory increments the data sequence, even if the header sequence (file allocation, protection, etc) remains unchanged. Below is a program that demonstrates the function. I've tested the program under Alpha VMS 7.2-1 and 7.3-1, the routine appears to have been introduced circa 1993. -- /* * Test for existence of f11x$posix_fastreadseqnos. This function returns * the sequence numbers held in the serialization lock for an open * directory, allowing you to quickly test for modification: * *hdrseqIncremented when header data changed (e.g. file * grows or shrinks). * *dataseq Incremented when file data changes (e.g. file renamed). * * Privileges: *Caller must have sysprv and cmkrnl privilege. * * Linking: *Link the the /sysexe qualifier (sys$system:sys.stb on vax) to resolve *the F11X$POSIX_FASTRDSEQNOS reference. * * Author: David Jones * Date:3-MAR-2003 * */ #include stdio.h #include stdlib.h #include string.h #include stat.h #include fibdef.h #include ssdef.h #include starlet.h #include descrip.h #include iodef.h struct seqno_st { long hdrseq;/* header data sequnece */ long dataseq; /* file data sequence */ }; /* * This function is part of the operating system image, link using * the /sysexe[/selective_search] qualifer. */ int F11X$POSIX_FASTRDSEQNOS ( short chan, struct seqno_st *seqno ); int main (int argc, char **argv) { int status; struct seqno_st seqno; short chan; struct stat info; static $DESCRIPTOR(devnam_dx,); struct { int length; struct fibdef *fib; } fib_desc; struct fibdef fib; struct { unsigned short status, count; long devdepend; } iosb; char line[80]; /* * Mislinking can cause system crash! do sanity check. */ if ( !F11X$POSIX_FASTRDSEQNOS ) { printf ( Program improperly built, must link /sysexe\n ); return SS$_ABORT; } /* * Stat the directory file named on the command line in order to * get device name and file id. */ if ( argc 2 ) { printf ( usage: fast_seqno filename\n ); return 1; } status = stat ( argv[1], info ); if ( status != 0 ) { printf ( Failed to stat file '%s'\n, argv[1] ); return 1; } /* * Assign channel and access file with speical fib$m_seqno that forces * XQP to keep an access lock on the file. */ devnam_dx.dsc$a_pointer = info.st_dev; devnam_dx.dsc$w_length = strlen ( devnam_dx.dsc$a_pointer ); status = SYS$ASSIGN ( devnam_dx, chan, 0, 0, 0 ); if ( (status1) == 0 ) { printf ( Assign to '%s' failed: %d\n, info.st_dev, status ); return status; } fib.fib$l_acctl = FIB$M_SEQNO | FIB$M_NORECORD | FIB$M_NOLOCK; fib.fib$w_fid[0] = info.st_ino[0]; fib.fib$w_fid[1] = info.st_ino[1]; fib.fib$w_fid[2] = info.st_ino[2]; fib.fib$w_did[0] = fib.fib$w_did[1] = fib.fib$w_did[2] = 0; fib_desc.length = 10; /* minimal FIB */ fib_desc.fib = fib; status = SYS$QIOW ( 0, chan, IO$_ACCESS|IO$M_ACCESS, iosb, 0, 0, fib_desc, 0, 0, 0, 0, 0 ); if ( (status1) == 1 ) status = iosb.status; if ( (status1) == 0 ) { printf ( Error in access, status: %d, iosb: %d %d %x\n, status, iosb.status, iosb.count, iosb.devdepend ); return status; } /* * Now enter loop where we fetch sequence numbers every time returnis * it. */ printf ( Each time you hit return, the current sequence numbers will\n); printf ( fetched and displayed. hit ctrl-Z to exit\n ); seqno.hdrseq = 0; seqno.dataseq = 0; while ( fgets ( line, 79, stdin ) ) { /* * XQP function must be called from kernel mode (probably * need pages locked as well). */ struct { long count; long channel; struct seqno_st *seqno; } arglist;
RE : Samba/Vms 2.2.7a
That will be corrected in the next release. -Message d'origine- De : Fezay Georges [mailto:[EMAIL PROTECTED] Envoyé : mercredi 26 février 2003 12:02 À : '[EMAIL PROTECTED]' Objet : Samba/Vms 2.2.7a Hello, I have installed Samba/Vms 2.2.7a with OpenVMS V7.2-1 and TCPIP V5.0A on an AlphaServer 800 5/533 . I have installed the last CRTL patch too. Everything seems to be OK except for the following : - i create a TXT file under OpenVMS with EDIT/TPU - i open it from Windows 2000 and i modify it (new lines of text) - i open it with OpenVMS editor : THE FILE IS EMPTY ! I have no problem if the file is created first with W2000 . The file record format when created with w2000 is : STREAM. The file record format when created with OVMS is : Variable Length. If i use set file toto.txt /rfm:stm , it is OK with OVMS but NOK with W2000 ! Did somebody encounter this problem ? Thanks __ / /\Georges Fezay (TRANSICIEL) /_/\_\ Administration Systeme OpenVMS/Windows NT - 2000 . \ \/ / VOLVO IT / IO \/_/tel: (33) 0472967075 - fax: (33) 0472967680 mailto:[EMAIL PROTECTED] ___ PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING: http://www.tuxedo.org/~esr/faqs/smart-questions.html
Samba/VMS 2.2.7a performances
Title: Samba/VMS 2.2.7a performances Hi. I am almost ready to give a new release of Samba 2.2.7a. It will be available later this week, or at least at the beginning of next week. The main change is that I completely re-wrote the TDB stuff (processing of the files who keep information about open sessions, locks on open files and other things). From the beginning, I had a lot of problems with the original code, due to the fact that it uses shared memory mapped files in a non VMS-compatible way. I got either corrupted files, or poor performances. So I eventually decided to do what I should have done from the beginning: forget about the mapped files and use indexed RMS files. Note that the drawback of the next release is that those of you who has joined a NT domain (smbpasswd -r xxx -j ddd) will have to do it again after installing the new release... The main purpose of this message is to give you some information about performances of this new release. I made some testing with Samba/VMS compared with Samba/Unix(tru64) and native Windows networking. The machines I used where very similar for VMS and Tru64, and a little less powerful for Windows. The VMS and Tru64 boxes are DS20E 666mhz, with efficient disks (HSZ70 controller with mirrorsets of RZ 4Mb disks). I started with the perl script that creates, reads and deletes a lot of files. The result is not so good. For 200 files, we get the following figures, in seconds, for respectively Windows, Tru64 and VMS : Create 9 10 46 Open/Close 4 3 4 Delete 3 5 21 Well, Samba/VMS is not very good at creating and deleting files, is it? That's not so surprising, because VMS file system is not too good at that either. The key point is that every changes in the VMS file system is immediately written back to disk (Unix or Windows don't do that: they keep everything in the file cache and actually write to disk later). However, in my opinion, creating hundreds of almost empty files and deleting them immediately after is hardly typical of a PC client activity, so I made some more testing, using ptime, a windows utility that can be found on the Net and gives the opportunity of measuring the time elapsed for running a command. I tried 3 actions : 1. Copy of a 8 Mbytes files from the local PC to the remote computer (Samba or other) 2. Open with Notepad of a 4 Mbytes file 3. dir of a directory containing 300 files Here are the results (again in seconds, with Windows, Tru64 Samba and VMS Samba): Copy 8Mo 3.9 8.0 5.5 dir 300 files 1.0 0.2 0.6/0.2 notepad 4Mb 4.0 5.8 3.0 Note that the 2 values for dir/VMS are for the first time and the subsequent times, showing the effectiveness of the memory caching process. For those tests, we can see that now, Samba/VMS is very similar, and sometimes better than Samba/Unix or native Windows (remember that my Windows testing box is less powerful than the tru64/VMS ones). To sum up, I think (and hope) that the upcoming release of Samba/VMS will be quite useable for standard users, most of the standard activity being done in an amount of time similar to other platforms (with the exception, anyway, of creating (copy to VMS) or deleting a great number of files, that remains significantly slower on VMS). I hope that you'll use and enjoy this new release. JY
RE : SAMBA v2.2.7a issues
Title: RE : SAMBA v2.2.7a issues Hi everyone. First, thanks for your thanks, and keep updated for the next release, which will be available within a couple of weeks, I hope, because it will be better... For the 1st question (Unknown socket option), I am sorry, but I did not remove from the valid list the IP options that are not available in VMS TCP/IP Services. I just left all the options that are defined in the standard Samba. Obviously, when you select an option which is not implemented by TCP/IP Services, you'll have this error message in the log. I'll try to think about removing what's not available in the next release. However (who knows?) may be those options will be available in a future release of VMS/TCP/IP ... For the 2nd question, I have no idea what could happen. I added more information in the error message (namely the buffer length, in order to know why it's invalid), so we'll be able to investigate further with the next release. JY Collot -Message d'origine- De : Tim Oakley [mailto:[EMAIL PROTECTED]] Envoyé : lundi 24 février 2003 16:44 À : [EMAIL PROTECTED] Objet : SAMBA v2.2.7a issues Hi everyone, I have recently installed the JYC SAMBA v227a and am really impressed by the performance enhancements (sincere congrats. to JYC). There are the usual teething problems but on the whole it is working fine. I do however have two specific questions to put to the community at present. 1. I have included the new smb.conf global option socket options = iptos_lowdelay In my client log files i see (very regularly) the message. Unknown socket option iptos_lowdelay If i run an smbtestparm the option is not flagged as unknown so why the message in the log files? 2. I have one particular user who has a repeated log file error message: receive_local_message. Error in recvfrom. (invalid buffer length). The error does not seem to create problems for the client but does mean that the log file for this client is very lengthy. Any suggestions ? Thanks. -- # Tim OAKLEY MAURY-IMPRIMEUR SA, Z.I, ROUTE D'ETAMPES 45330 MALESHERBES France Voice: (+33) 02.38.32.34.38 Fax: (+33) 02.38.32.37.72 Email : [EMAIL PROTECTED]
RE : Problem file lookup on 2.2.7a Alpha
Title: RE : Problem file lookup on 2.2.7a Alpha Hi. Well, you are perfectly right, and it actually is an issue... For performance reasons, I try to cache data in memory (directory contents and file characteristics), and only go back to actually reading the disk only when needed. The problem is that I consider that the disk has changed only when some internal VMS structure (namely the Volume Lock Value Block) has changed. As some of you may have witnessed, this gives really a lot of enhancements... Unfortunately, this data changes only when there is either allocation or deallocation on the volume, so if you just rename a file from outside your Samba session, the number of free blocks does not change, so it's not seen in your session before someone either creates or deletes something (changes the number of free blocks on the volume, anyway), or if you do something within your Samba session that forces to invalidate the caches (such as renaming something, creating or deleting a file, ...). I don't know what I can do about that, except removing the memory caching (and then multiply the browsing time by may be 30 or 40)... JY Collot -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] Envoyé : lundi 24 février 2003 17:11 À : [EMAIL PROTECTED] Objet : Problem file lookup on 2.2.7a Alpha Hi evereybody, I successfully installed Samba 2.2.7a on an AlphaServer GS140 running OpenVMS 7.2-2 All things I've tested look fine, except one problem : First I connect my homeshare. Looks great, I can see all my files. Then, I rename a file on OpenVMS for example $ RENAME TEST.TXT TEST1.TXT Then I refresh the view from the Windows explorer and nothing changes. No new file test1.txt appears. Instead I still see a file named test.txt. When I now try to open this file, a message says, that this file can not be found. Can anyone reproduce this problem ? or do I have to change some options ?!?! Have a nice day Martin Philippi
RE : RE : Problem file lookup on 2.2.7a Alpha
Title: RE : RE : Problem file lookup on 2.2.7a Alpha No, as far as I know. -Message d'origine- De : Michael D. Ober [mailto:[EMAIL PROTECTED]] Envoyé : lundi 24 février 2003 17:52 À : 'COLLOT Jean-Yves'; '[EMAIL PROTECTED]' Cc : [EMAIL PROTECTED] Objet : RE: RE : Problem file lookup on 2.2.7a Alpha Does VMS have a function similar to the Windows WaitForFileSystemChange? Mike. -Original Message- From: COLLOT Jean-Yves [mailto:[EMAIL PROTECTED]] Sent: Monday, February 24, 2003 9:48 AM To: '[EMAIL PROTECTED]' Cc: [EMAIL PROTECTED] Subject: RE : Problem file lookup on 2.2.7a Alpha Hi. Well, you are perfectly right, and it actually is an issue... For performance reasons, I try to cache data in memory (directory contents and file characteristics), and only go back to actually reading the disk only when needed. The problem is that I consider that the disk has changed only when some internal VMS structure (namely the Volume Lock Value Block) has changed. As some of you may have witnessed, this gives really a lot of enhancements... Unfortunately, this data changes only when there is either allocation or deallocation on the volume, so if you just rename a file from outside your Samba session, the number of free blocks does not change, so it's not seen in your session before someone either creates or deletes something (changes the number of free blocks on the volume, anyway), or if you do something within your Samba session that forces to invalidate the caches (such as renaming something, creating or deleting a file, ...). I don't know what I can do about that, except removing the memory caching (and then multiply the browsing time by may be 30 or 40)... JY Collot -Message d'origine- De : [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] Envoyé : lundi 24 février 2003 17:11 À : [EMAIL PROTECTED] Objet : Problem file lookup on 2.2.7a Alpha Hi evereybody, I successfully installed Samba 2.2.7a on an AlphaServer GS140 running OpenVMS 7.2-2 All things I've tested look fine, except one problem : First I connect my homeshare. Looks great, I can see all my files. Then, I rename a file on OpenVMS for example $ RENAME TEST.TXT TEST1.TXT Then I refresh the view from the Windows explorer and nothing changes. No new file test1.txt appears. Instead I still see a file named test.txt. When I now try to open this file, a message says, that this file can not be found. Can anyone reproduce this problem ? or do I have to change some options ?!?! Have a nice day Martin Philippi
RE : Samba/Vms 2.2.7a ODS-5 support
Title: RE : Samba/Vms 2.2.7a ODS-5 support Sorry about that, sometimes my guesses are wrong... However, you should test this problem with the new version, because I don't encounter it myself. JY -Message d'origine- De : Carl Perkins [mailto:[EMAIL PROTECTED]] Envoyé : jeudi 23 janvier 2003 13:46 À : [EMAIL PROTECTED] Objet : Re: Samba/Vms 2.2.7a ODS-5 support For example, I suspect that the Excel save as... and New Excel... = problem that was reported a couple of days ago by Carl Perkins should be = fixed with this release (providing that Mr Perkins uses ODS-5, which I = guess is true). Jean-Yves Collot Actually, I don't. These problems were on an ODS-2 volume. --- Carl
RE : Samba/Vms 2.2.7a ODS-5 support
Title: RE : Samba/Vms 2.2.7a ODS-5 support Yes. -Message d'origine- De : Michael D. Ober [mailto:[EMAIL PROTECTED]] Envoyé : jeudi 23 janvier 2003 15:33 À : Samba VMS Objet : RE: Samba/Vms 2.2.7a ODS-5 support Does this version also fix the Win98 name mangling issue? QUOTE From Jan-erik Söderholm again : A problem with directories with more than 8 characters and Notepad/Wordpad with Win98 client. That's very true too. It's due to one of my optimizations, which was not so good after all, and which is now useless anyway. I was not aware of this bad effect because I have no Win98 clients. It will be fixed in the next release. /QUOTE Thanks for all your hard work on SAMBA VMS. Mike Ober. -Original Message- From: Jean-Yves COLLOT [mailto:[EMAIL PROTECTED]] Sent: Thursday, January 23, 2003 4:26 AM To: Samba VMS Subject: Samba/Vms 2.2.7a ODS-5 support Hi. There is a new release of Samba/VMS 2.2.7a at the usual location http://www.pi-net.dyndns.org/anonymous/jyc/ This release supports ODS-5 file system, if (and only if...) you run VMS Version 7.2-1 or later, and you have downloaded and installed the latest CRTL patches for your version, which you can find at http://ftp.support.compaq.com/patches/.new/openvms.shtml This release should fix most problems you may encounter with ODS-5. For example, I suspect that the Excel save as... and New Excel... problem that was reported a couple of days ago by Carl Perkins should be fixed with this release (providing that Mr Perkins uses ODS-5, which I guess is true). Enjoy... Jean-Yves Collot