Re: [U2] [UV] Type 30 Semaphore Stuck - HELP!!!
Scott,
If you have UVNET, there is a work around that I came up with: Create
a shadow account with all file pointers that go to dynamically
hashed files repointed to the network version of the file name, ie
myserver!/diskpath. Then repoint your UniObjects to the shadow
account.
The UVNET version of the rpc mechanism does not have the same problem
closing dynamic files as UniObjects.
-Rick
On Apr 30, 2009, at 3:43 PM, iggch...@comcast.net wrote:
Thanks for the info Rick.B I'll give it a look.
- Original Message -
From: Richard Nuckolls r...@lynden.com
To: u2-users@listserver.u2ug.org
Sent: Thursday, April 30, 2009 11:35:49 AM GMT -06:00 US/Canada
Central
Subject: Re: [U2] [UV] Type 30 Semaphore Stuck - HELP!!!
If, by chance, you are using UniObjects, there is a bug in it, fixed B
at 10.2.22, that can cause this problem when a session times out or B
otherwise exits in a less than optimal manner.
I think that a corrupt file or index can also cause this problem
Alternately, you might check to see if there is anything in {uvhome}/
errlog.
Rick Nuckolls
Lynden Inc
On Apr 29, 2009, at 11:38 AM, iggch...@comcast.net wrote:
Hi All,
UV10.2.7 Pick
HPUX 11.11
Someone (ok me) decided to resize a file today from dynamic to B
static.B
ShortlyB a fter that, the system came to a halt because the type 30 B
semaphore
flag was stuck.B B Can anyone tell me how/why this happened? The
file B
I resized
is fine.B B I've been using the CONCURRENT flag with the resize B
lately B and
haven't encountered any issues.B B However, I've never (and will B
never again)
change from dynamic to static with otherB usersB on the system. B B
I B
did clear
the lock (UNOCK T30LOCK SEMAPHORE) but about an hour later it got B
stuck
again.B B I found the users that had the file open using /etc/fuser B
pathname
and had them log out.B B Is there anythingB B I can check to find
out B
why this
keeps happening?
Also, any inside knowledge on how the resize using the concurrent B
option
actually works would be appreciated.
Thanks muchB in advance,
Scott
---
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
---
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
---
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
---
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] UniObjects for Java on Ubuntu anyone?
I have not used Ubuntu, but try optioning Java with - Dfile.encoding=iso8859-1 If you are were tomcat, this would go in the CATALINA_OPTS environment variable. (setenv.sh) -Rick On Feb 13, 2009, at 2:34 PM, Jeff Powell wrote: The LANG variable issue takes a new twist with Ubuntu Server 8.10. Neither LANG=C (IBM recommended) or LANG=en_US.iso885915 (must use for RedHat) will work. The value marks are wrongly used as field marks. And in case you're wondering no, LANG=en_US.UTF-8 also does not work. Has anyone made UniObjects for java work on Ubuntu? Thanks. Jeff --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Indexes have disappeared
Are the file systems truly identical, or are you using symbolic links to make them look the same? Use SET.INDEX filename INFORM to see where your index is really pointing. -Rick On Apr 22, 2008, at 3:12 PM, Louie Bergsagel wrote: We use flashcopy to move data from one computer to another, and recently noticed that UniVerse is no longer recognizing the indexes even though they are copied along with the files. We recently upgraded to UniVerse 10.2.7 (Pick syntax) Has anybody heard of this weirdness? -- Louie in Seattle --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] UV Autologout Quirks
I would guess that the result of autologout on Unix or Linux may depend on how Universe was invoked. If it was invoked from a Unix shell, such as ksh, as a command then autologout probably just exits Universe, leaving the user back in the shell. On the other hand, if the original .profile, or a script masking the uv command uses exec uv to start Universe instead, then the user should be completely out of the system after being logged out of Universe. -Rick Nuckolls Lynden Inc. On Feb 12, 2008, at 7:26 AM, Kevin King wrote: A typical user with a 30 minute logout time shows up to 14+ hours of idle (via who -u in Linux) and yet according to their autologout time they should have been gone hours earlier. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] UniObjects error 30105 Unable to load subroutine on server
Okay, but do you reset the routine name before each call? This should not be necessary, but I think that that is what the error message is implying is wrong. -Rick Nuckolls On Feb 2, 2008, at 2:44 PM, Anthony W. Youngman wrote: And all the arguments are assigned in my VB app before I call it --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Uniobjects for Java
You need to set your classpath, which indicates where you need to look for code. In your case, were you are probably executing javac in the same location that you have placed the test program, try set CLASSPATH=C:\IBM\UniDK\uojsdk\asjava.zip;. Note the period after the semicolon which indicates that you are including the current directory. See http://java.sun.com/j2se/1.3/docs/tooldocs/win32/classpath.html for a write up. -Rick = Richard Nuckolls Lynden, Inc. On Jan 22, 2008, at 8:34 AM, [EMAIL PROTECTED] wrote: I am checking out Uniobjects for Java and am putting together a test app. I include the line import asjava.uniobjects.*; outside of the Class, but get a message The import asjava cannot be resolved. The Uniobjects software from the client disc is installed at C:\IBM\UniDK\uojsdk. Do I need to register the library, or provide the full path to the lib? Charles Shaffer Senior Analyst NTN-Bower Corporation --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Universe usa hangs
This is a known bug in 10.2.x The solution is to get the 10.2.0 version of the usa/usm/usp binary. -Rick = Richard Nuckolls Lynden Inc. On Jan 11, 2008, at 6:38 AM, Kevin King wrote: I'm assisting with getting UV10.2 up on a SuSE 10.1 box, and there's a couple of things that are disturbing me. First, the /usr/bin/disable command doesn't seem to want to disable a print queue, but more importantly when issuing this command from bash: usa -p printer -F form ...where printer and form are a valid printer and form name, usa does not return to the $ prompt. If you press Ctrl-C and check the printer via: usa -p printer ...it does in fact show the form name as mounted on the printer. But I would think this should not require a Ctrl-C. Any clues? -Kevin --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] IBM web site
I cannot say that it is on the front page, but I had no trouble finding it in menus at the ProductsSoftwareSoftware by Catagory Data Database Management Tools Page. :-D -Rick Nuckolls Lynden Inc On Dec 20, 2007, at 8:00 AM, Martin Phillips wrote: Hopefully nothing sinister here but has anyone else noticed that the U2 products appear to have vanished from IBM's product list on their web site. Support and training are still available but I cannot find any reference in the areas one would visit as a prospective purchaser. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] locked file help in U2 VER 9 NT 4
I have seen this happen when a file gets an inconsistent state flag set. Are you doing Transaction Logging? Try RECOVERY.CONSISTENT pathname to reset the flag. = Richard Nuckolls Lynden Inc. [EMAIL PROTECTED] On Nov 30, 2007, at 12:46 AM, Dave R wrote: My inventory file is locked and will not allow updates (writes or ed/fi) it hangs and then logs me out. I looked at UVFIXFILE but was unsure of what flags to run. I tried CLEAR ALL that cleared 2 locks but didn't release the file. I rebooted and that server and still no release. I then created a bu-inv file and copied the records into the new file. I tried in windows to rename the old file and windows said there is a sharing violation. I eventual set the Voc to point to the back up file so every one could go back to work. I am will shut down the Universe services this week end and try to rectify the problem un less some has a hot solution. Any suggestions would be usefully. Dave R e Fax (815)4259364 P.O. Box 17811, Irvine CA 92623-7811 --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Putting Universe on a SAN
re SAN's and Universe: DELETE.FILE and RESIZE.FILE will not work on dynamic files that are NFS mounted. (Universe keeps the DATA.30 file open while trying to delete the enclosing directory. This will fail on an nfs mounted directory.) Override the VOC entry with your own version. Do not access a snapshot of a file from the same mount as the original. At least on NetApps, the inode of the snapshot will be the same as the inode of the original. Universe will confuse the data. Besides the usual caveats about making sure that a single machine controls locks for any given file, it is also worthwhile to keep the NFS under a set folders(4-6 or so) and then give each of these folders its own mount point on the host machine. Effectively creating multiple devices under UV. Of course: do not try this without a good UPS. Most SAN's have one built in too. Corruption of a group is still possible, but our experience has been good. My advice would be to replicate snapshots rather than live data (Unless you are using Uv Replication). With snapshots, you can temporarily suspend IO, take a snapshot, and continue. If you are doing live replication at the SAN level, then a major problem (an errant DELETE.FILE, for instance) will be replicated on your backup SAN. Also, suspending IO will avoid catching files in transitional states. Good Luck! -Rick Nuckolls Lynden, Inc On Nov 9, 2007, at 11:01 AM, Jerry Banker wrote: Does anyone know if the problems they were having were fixed or could be fixed? Not at 10.2.2 for those mentioned above. I have not used Uv Data Replication. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] [UV] Installing UniObjects
The unirpcd is the only thing that is necessary on the server side, and this comes with 10 above. asjava.zip needs to be in the classpath on the client. You will ultimately be better off, I think, if you do two things: Create wrapper classes around the primary asjava classes so that you can modify their behavior. Plan on using middleware, such as servlets, for actually accessing Universe in a production situation. The list could extend for quite a ways. There are enough booby traps there to keep you entertained for a while. More advice: Use subroutines as much as reasonably possible. Create a UvSession interface, and program to it rather than to UniSession. This will help in many ways in the long run. Use a UV subroutine that opens files to common on the back end and hands out the index to the location. (You can only open a small, arbitrary number of UniFiles simultaneously, so the common will allow you to extend that number.) Try to do as much conversion as you can on the host, prior to moving data to the client. The asjava routines do conversions by calling back to the host, which is terrible inefficient. (Or, if you are slightly nuts, you can re-implement the common conversions in Java. Hmmm, maybe I was nuts a while back.) The innards of UniObjects can be confusing at points. If you try to access UV delimited strings outside of the UniDynArray class, you will eventually stumble on character encoding issues, wherein the delimiters are improperly mapped. The problem is that some platforms default to UTF-8 for converting between bytes and characters. The asjava classes could easily deal with this, but they have not. This can be mitigated by using the following option on the Java executable: -Dfile.encoding=iso8859-1 There was an issue, #8776, that was finally fixed in 10.1.22 ( I think), and also in 10.2.1, that can affect Universe adversely if a client session times out or otherwise terminates abruptly. If you are running any client software (UCI, UniObjects for Java, or the other kind), and use dynamic files, then move to a version without this issue. Dangling group locks can be irksome. Good luck! -Rick On Nov 8, 2007, at 6:43 PM, Ray Wurlod wrote: As far as I'm aware (in version 10) it's just there - you don't need to install anything. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] UV - 10.2.2 [not-secure]
Clifton, usa, usm, usp (all the same executable) hang when trying certain operations. They problem evidently started with 10.2.1, and fixed at 10.2.6 (not available for Solaris yet) usa -o -pr0 for instance. Note that uvadmin, the admin menu, etc all execute these and will hang when disabling printers or changing forms. They actually take the appropriate action, it is just that you have to break or or kill the task. The IBM solution seems to be to wait until you open a case, and then send out a copy of usa rev 10.2.0, which works with 10.2.2 usd. I wasted a few hours trying to figure out what I'd done wrong with the spooler in our upgrade. The release notes for 10.2.6 are available for some platforms from the TechConnect availability matrix. Too bad I did not think of that first. -Rick Nuckolls Lynden, Inc On Nov 6, 2007, at 1:41 PM, Clifton Oliver wrote: Mark, Would you mind giving just a brief description of what you mean by broken please? Not in detail (unless you want to grin) but enough to let the rest of us know what to watch out for. Thanks. On Nov 6, 2007, at 12:43 , Hennessey, Mark F. wrote: Also, usa is broken at 10.2.2 --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] UV - 10.2.2 [not-secure]
You will need to check with IBM for the answer, but I suspect that it is true on a number of platforms. Who knows about Windows. It is definitely true on Solaris. -Rick On Nov 6, 2007, at 4:49 PM, Clifton Oliver wrote: I am interpreting that to mean this is on those releases on all platforms (AIX) not just Solaris, correct? --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] curious EQUATE issue - SOLVED Just want to know if anyone understands WHY?
regarding the example below:
EQU VALID.CC.TYPES TO CHANGE('A.B.D.M.S.V','.',@VM)
The problem with this is that the CHANGE within the EQU will get
evaluated at run time, so it will be done repeatedly if placed within
a LOOP or FOR structure.
You would be better off with the earlier suggestion of
EQU VALID.CC.TYPES TO ('A':@VM:'B':@VM:'D':@VM:'M':@VM:'S':@VM:'V')
The concatenation will be done by the compiler, and the parenthesis
will protect the LOCATE field designator, 1,1 from being
interpreted as an extraction. This does compile, by the way.
-Rick
On Oct 23, 2007, at 2:04 PM, Rex Gozar wrote:
Allen,
EQU VALID.CC.TYPES TO CHANGE('A.B.D.M.S.V','.',@VM)
does compile and it will protect the value of VALID.CC.TYPES from
being unintentionally changed -- which is the reason for using EQU
to declare constants in the code.
I just wanted to remind people that equating to a variable does not
set an unchanging, constant value; this can be a source of hard-to-
find bugs.
rex
---
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
---
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] curious EQUATE issue - SOLVED Just want to know if anyone understands WHY?
When the new version of the equate was compiled under Information flavor, the compiler warned that it was a Reality style LOCATE. To its credit, the compiler defers the evaluation of the EQU until it is used, at which point, it evaluates it in context, so the compiler is evaluating VALID.CC.TYPES1,1 as A':@VM:'B':@VM:'D':@VM:'M':@VM:'S':@VM:'V'1,1 The vlist output below gives a good hint as to what is going on. Notice that the field,value is being applied to the last value of the EQU string prior to doing the build with the rest of the codes. In the Reality format, it is legitimate to omit the field,value specifier, in which case the program will do a field level search rather than a value level search. This quirk would only show up with the use of an EQUATE rather than a variable assignment. If you feel compelled to use an EQUATE for the codes, you could overcome the issue with a temporary assignment such as EQU VALID.CC.TYPES TO 'A':@VM:'B':@VM:'D':@VM:'M':@VM:'S':@VM:'V' ... CC.TYPES = VALID.CC.TYPES LOCATE X IN CC.TYPES1,1 ... - VLIST of LOCATE on original equate -- 5: LOCATE X IN VALID.CC.TYPES1,1 BY 'AL' SETTING LP THEN Press any key to continue... 5 E : 060 dyn_extractV 1 1 0 = $R0 5 0001A : 03A concat A}B}D}M}S} $R0 = $R1 5 00022 : 0D4 locate X $R1 1 0 0 AL = LP 5 00032 : 0C4 jumpf --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] How does one safely validate that a session is connected in UniObjects (VB VB.Net)
The same result can be accomplished by calling a subroutine that does nothing other than return. If it does not throw an exception, then the connection is alive. == Richard Nuckolls Lynden Incorporated 18000 International Blvd, #601 Seattle, WA 98188 On Oct 4, 2007, at 8:42 AM, Kryka, Richard wrote: Jef Lee wrote: I believe the IsActive property is not entirely accurate... From my tests, you are right about IsActive - it is the results of the last connection (I think) and has nothing to do with the current state. I added a timer function to our VB front end that runs every 60 seconds and just opens a file. If the connection has been lost, the open is unsuccessful. We display a message informing the user that the connection is broken and they need to log on again. This function has almost no overhead and keeps users from spending several minutes or more on a form, then discovering that nothing was saved. It also solved the timeout issue where people would go into a meeting or out to lunch and their connection would be timed out for inactivity. The function runs from only from 6am to 8pm, their local time. If users go home at night without logging out, their connection is still logged out for inactivity instead of being connected all night. Dick Kryka Director of Applications Paragon Financial Services a Division of Money Management International 303-632-2226 [EMAIL PROTECTED] --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] [UV] Weird SEARCH Results
Consider looking at that line in control-character display mode (enter ^ cr at the ED : prompt) -Rick On Jul 31, 2007, at 8:37 AM, [EMAIL PROTECTED] wrote: Hi all, uv10.1.8 - Pick Format hpux 11 Anyone have an idea on what might be wrong here? SEARCH ROM.BP STRING:OES210 STRING: 0 record(s) selected to SELECT list #0. ED ROM.BP OEM110 1806 lines long. : L OES210 0374: CALL OES210(SOD.FILE,IM.FILE,SOH.ID,MAT DETAILS) It seems to be limited to the one program (OEM110) as this program is never returned in any search results. Thanks in advance. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] UniSubroutine error
The intercall.pdf guide defines it as IE_DATUMERROR Internal Datum Error. A little vague, but it probably indicates that some data failed to convert. It is most heavily referenced in the uvodbc.pdf document. -Rick On Jul 31, 2007, at 8:26 AM, Jeff Powell wrote: Does anyone know what 30102 (unknown) code means? Thanks. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Differences between UV and UD
Okay, I have been sucked into this conversation: I would also suggest that if the UniData folks were satisfied with PI/PI Open/ UniVerse, they would not have gone to the trouble of creating UniData. The early versions of Unidata were poor cousins to PI/Open (imho). One of the Unidata engineers sat in our offices many years ago and said that their versions of Basic and the query language were based directly on the Prime Information manuals. ( They did not have a copy of PI in the office while they were working to emulate it.) My understanding is that Unidata had a database, and was looking for an established market that they could enter with it; they chose the MV market. Vmark (which predated Unidata by a number of years, I believe) saw an opportunity to move the MV market from Reality and Primos to Unix, which they correctly perceived as supplanting the existing proprietary minicomputers of the time. It took several years after Vmark released Universe for Prime to release their (excellent) Unix version of Prime-Information. Prime failed for familiar reasons: Their development and marketing structure was based on the high-margins of proprietary systems, and they were the target of a prolonged hostile takeover attempt, which dissuaded customers from updating their equipment and cost the company the farm. In retrospect, Unix had probably doomed them; I doubt that they could have shifted their structure fast enough to be successful with it. :-) --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Differences between UV and UD
Google sez: Vmark was founded in 1984; Unidata in 1986. I did not find a date for PI/Open. If I remember correctly Vmark built the first releases of PI on Unix for Prime. Highly doubtful. Vmark was a direct competitor of Prime. The original Prime Information was written by Devcom in Bellevue, Wa, in the very late '70's. Substantial portions were written in PMA (Prime macro-assembler). Most of the rest was written in Fortran or Bootstrapped in Basic. PI/Open was a fresh start, written by a collection of Brit's, in large part, who did a good job of separating/abstracting the function of the product from the OS that it was running on. They rewrote the product from scratch in C. Prime was parted out in 1988, and I think that may have been when Vmark bought PI/Open. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: Spam:Re: [U2] Differences between UV and UD
Thanks Lance, Ah, I had forgotten about PI/EXL. I could not believe that Universe had anything to do with the core of PI/Open. -Rick On Apr 30, 2007, at 2:17 PM, Lance J. Andersen wrote: We rebranded a a version of uniVerse as PI/EXL prior to the release of PI/Open. PI/Open, was *not* from vmark, it was created internally at Prime, with most of the work done at the Prime UK facility. Martin Phillips was part of the UK team. -lance Jerry wrote: If I remember correctly Vmark built the first releases of PI on Unix for Prime. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Connecting to a remote Universe server in Universe Basic
On Apr 30, 2007, at 4:56 PM, Jason Lin wrote: That may not be an issue if all we need to do is query data on the remote server. Even if only one system is doing updating, locks are an issue for (at least) two reasons: The first is that that a group's structure (or for dynamic files, the hashing modulus) may appear to be invalid to a remote system that is reading an NFS mounted file that is concurrently being modified by the system that owns the file. A more dangerous problem may occur with dynamic files because information from the file header is cached when a dynamic file is opened. For the remote system, this copy will become out of date as the file is updated by the directly connected system. There is also the possibility that the file header will overwritten by the out of date version when the remotely mounted version is closed by Universe. Uv/Net is a separately priced product but probably worth it for your situation. You only need it on the system that is accepting the remote access, not the one that is initiating them. -Rick --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] RESIZE Error
Windows or Unix? Universe or Unidata? NFS mounted file system? On Apr 24, 2007, at 9:56 AM, Baruch Salamander wrote: I'm trying to RESIZE a file, but I'm getting a RESIZE error. RESIZE: Error on [filename]. File not resized. As a follow-up, an attempt was made to copy the content to a temporary file, delete the old one, re-create it, and copy the content back, but the system refuses to delete the old one due to this error. What's the best way to get around this so that the file gets properly resized? Baruch --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] time and time again.
Hard to say for sure, but... I do not think that there is a direct patch for MS for NT4 and the timezone thing. However, you can us an MS utility, tzedit.exe to set the Pacific tz to change on the second Sunday of March, and reverse on the first Sunday of November. But in order to make Windoz happy with the fix, you need to select a different tz, click apply, and then reselect the Pacific tz and apply again. Then be sure that your Windows time is correct and has the timezone box checked. Good luck. -Rick On Apr 5, 2007, at 1:13 AM, Dave R wrote: May be someone can explain how to get back on the correct time. Paragraph time and system time are 1 hr apart. This started this week after the normal DLS time and it is causing my sleep process to start 1hr later then they should. I am running NT4 and UniVers 9.4 (no the company doesn't want to Upgrade we are switching to Oracle. :-) .-) .-) TIME 16:59:47 04 APR 2007 02 P.TIME (paragraph) 16:59:36 04 APR 2007 02 DR.TIME (Oconv(TIME(),'MTS')) 17:59:38 64778.671 (time()) P.S. I just rebooted the system and the problem is still there and is set to DLS. Dave R 949 575 8519 e Fax (815)4259364 P.O. Box 17811, Irvine CA 92623-7811 --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: [U2] Dynamic files, big transactions
Andre, I am not quite sure what you are doing a transaction on, but even if you successfully configured a Universe system to handle transactions with that many elements simultaneously locked, I would guess that it is inefficient to grab that many locks at once. Your transaction is too (bloody) large! So I think that you may need to restructure your process somehow. The first question that I would have is are you doing a BASIC transaction or a SQL one? I am assuming that you are doing a BASIC one. Beyond that, you may need to NOT use a formal TRANSACTION, but rather keep track of the state of your update somewhere, and do WRITES outside of a TRANSACTION block. If you are just creating new records, and you ensure that the id is unique, you could save some thrashing by doing WRITES into a file without a previous READU. One scheme might be: READU the sequence counter record, increment by the number of detail records in the transaction. Create a transaction control record (TCR) in a file that you have created for this purpose. perhaps a usernumber together with time() could be used as an id. Include the range of detail ids in the TCR WRITE (and unlock) the sequence counter. Create the detail records in a work file (without using transactions, and possible without locking the records before writing) Update the TCR, indicating that the previous step completed Copy the records for the TCR's range from the work file to the production file, deleting them out of the work file. stamp the TCR as complete. --- This is somewhat crude and incomplete, but you get the idea. Out of curiosity, what kind of event has that many elements? Good luck! -Rick Nuckolls Lynden, Inc. On Mar 12, 2007, at 8:10 AM, Andre Meij wrote: All, I have done some more digging and I'm now stuck at writing about 6k new records within one transaction. I don't know what I might be missing but depending on different variations I get a max of 5k or 6k and then a READU threshold reached error will break the transaction. Any more insights will be greatly appreciated. I have pasted info of my last test below. Regards, Andre Meij Innovate-IT DELETE.FILE TEST06 DELETEd TEST06, Type 30. DELETEd file D_TEST06, Type 3, Modulo 1. DELETEd file definition record TEST06 in the VOC file. CREATE.FILE TEST06 30 MINIMUM.MODULUS 1024 SEQ.NUM Creating file TEST06 as Type 30. Creating file D_TEST06 as Type 3, Modulo 1, Separation 2. Added @ID, the default record for RetrieVe, to D_TEST06. ANALYZE.FILE TEST06 File name .. TEST06 Pathname ... TEST06 File type .. DYNAMIC Hashing Algorithm .. SEQUENTIAL No. of groups (modulus) 1024 current ( minimum 1024 ) Large record size .. 1628 bytes Group size . 2048 bytes Load factors ... 80% (split), 50% (merge) and 0% (actual) Total size . 2101248 bytes RUN BP test 15000 5 Starting writing 5 records READU threshold reached, lock on 11521224 denied! Completed writing 6223 records Error: WARNING, 'TEST06' record '11521224' locked, try again later. GROUP.STAT.DETAIL TEST06 Type description= Hashed, keys end in numbers. Bytes Record.id File= TEST06 Modulo= 1024 Sep= 4 Type= 30 32 11518037 32 11520098 32 11519718 32 11516722 32 11517008 -- 160 Bytes 5 Records in Group 1 32 11518036 32 11517009 32 11520099 32 11519719 32 11516723 -- 160 Bytes 5 Records in Group 2 32 11518038 32 11517011 32 11520096 32 11519716 32 11516720 LIST.READU (started before the end of the transaction after the READU error. Active Record Locks: Device Inode Netnode Userno LmodePid Login Id Item-ID. 22282244 12580680 115 RU 21074 ameij 11515011 22282244 12580680 115 RU 21074 ameij 11515033 22282244 12580680 115 RU 21074 ameij 11515068 22282244 12580680 115 RU 21074 ameij 11515102 22282244 12580680 115 RU 21074 ameij 11515135 22282244 12580680 115 RU 21074 ameij 11515146 22282244 12580680 115 RU 21074 ameij 11515161 22282244 12580680 115 RU 21074 ameij 11515179 22282244 12580680 115 RU 21074 ameij 11515218 22282244 12580680 115 RU 21074 ameij 11515235 22282244 12580680 115 RU 21074 ameij 11515284 22282244 12580680 115 RU 21074 ameij 11515298 22282244 12580680 115 RU 21074 ameij 11515315 22282244 12580680 115 RU 21074 ameij 11515334 22282244 12580680 115 RU
Re: [U2] Dynamic files, big transactions
Andre, This seemed very strange, since normally 250 record keys would never hash into the same group of a dynamic file. The exception might be if you created a new part file and then sought to lock and add a large group of records at one time. The new dynamic part file would have a modulus of 1, and all of the records that hashed into that file would be locked in the same group until some of them were written, and the file split to a larger modulus. If your algorithm for the Distributed file could cause this situation, then the solution may be to create the new parts with a MINIMUM.MODULUS value large enough to split out the record keys into separate groups (~23? Bigger is better). CONFIGURE.FILE MINIMUM.MODULUS ... (in the Prime/PI Open version) would accept a keyword of IMMEDIATE to force the splitting of groups. Universe lacks this option, so you should specify MINIMUM.MODULUS at the time that you create the part file. -Rick Nuckolls Lynden Inc. On Mar 9, 2007, at 10:39 AM, Andre Meij wrote: Hi, We have a highly technical problem with universe related to the locking tables and their configuration: We have a big application running on Universe 10.1 (Solaris). This application is build on Distributed Dynamic files. Some of the keys are auto numbers; others are supplied by external systems. With our current settings we can have a maximum of 250 locks in one group; this means most certainly for auto numbers we can only lock records 1000 to 1249 without getting an abort and a rollback because a new lock cannot be acquired. This 250th lock cannot be acquired because all these locks fall within the same lock group which is limited to 250 locks. I know of 2 uvconfig parameters that define this locking behavior however the maximum settings of these settings are limited by the maximum size of the shared memory segment. # GLTABSZ - sets the number of group lock entries GLTABSZ 250 # RLTABSZ - sets the number of read lock entries RLTABSZ 250 Our testing indicated that these numbers cannot be raised any higher (due to the shared memory limit). This all means that I cannot lock more than 250 records in one transaction; this is unfortunately not always enough, we occasionally have to implement some extensive tricks to circumvent this. I very much would like to see this resolved on a Universe level so that the programmers can stop worrying about this. Anyone who has experience or knows of someone with experience, please help :-). Maybe you have knowledge of this problem yourself or know of somebody within IBM who could help us resolve this. Regards, Andre Meij Innovate-IT --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
Re: Spam:Re: [U2] Dynamic files, big transactions
Jerry, Observe: In the example below, you will note that CONFIGURE.FILE has no effect on the physical size of the ALPHA file, and the size is defined by the current modulus. The modulus can be computed from the size of a dynamic file using modulus = (totalbytes - 2048)/2048, which will yield 23 for the BETA file size. To double check that there was not some sort of sparse creation algorithm involved, I added 3 small records to the modulus 23 ALPHA file, and found that they all appeared in the same group. ANALYZE.FILE will give you the true modulus (see below). In the case of ALPHA, deleting records will not contract the file unless the current modulus is greater than 23. These examples are on a Solaris system. There is a chance that Windows behaves differently. -Rick :CREATE.FILE ALPHA DYNAMIC Creating file ALPHA as Type 30. Creating file D_ALPHA as Type 3, Modulo 1, Separation 2. Added @ID, the default record for RetrieVe, to D_ALPHA. :! ls -l ALPHA total 16 -rw-rw-r-- 1 rick piadm 4096 Mar 9 13:53 DATA.30 -rw-rw-r-- 1 rick piadm 2048 Mar 9 13:53 OVER.30 :CONFIGURE.FILE ALPHA MINIMUM.MODULUS 23 File name = ALPHA Changing Minimum Modulus from 1 to 23. File name = ALPHA configured. :.X2 02 ! ls -l ALPHA total 16 -rw-rw-r-- 1 rick piadm 4096 Mar 9 13:53 DATA.30 -rw-rw-r-- 1 rick piadm 2048 Mar 9 13:53 OVER.30 :CREATE.FILE BETA DYNAMIC MINIMUM.MODULUS 23 Creating file BETA as Type 30. Creating file D_BETA as Type 3, Modulo 1, Separation 2. Added @ID, the default record for RetrieVe, to D_BETA. :! ls -l BETA total 112 -rw-rw-r-- 1 rick piadm 49152 Mar 9 13:54 DATA.30 -rw-rw-r-- 1 rick piadm 2048 Mar 9 13:54 OVER.30 : :ANALYZE.FILE ALPHA File name .. ALPHA Pathname ... ALPHA File type .. DYNAMIC Hashing Algorithm .. GENERAL No. of groups (modulus) 1 current ( minimum 23 ) Large record size .. 1628 bytes Group size . 2048 bytes Load factors ... 80% (split), 50% (merge) and 2% (actual) Total size . 6144 bytes On Mar 9, 2007, at 1:14 PM, Jerry wrote: Did you mean that UniVerse doesn't have the CONFIGURE.FILE or what? Because UniVerse has the CONFIGURE.FILE command, the keyword MINIMUM.MODULUS, and it works immediate without using the IMMEDIATE key word. - Original Message - From: Richard Nuckolls [EMAIL PROTECTED] To: u2-users@listserver.u2ug.org Sent: Friday, March 09, 2007 2:35 PM Subject: Spam:Re: [U2] Dynamic files, big transactions CONFIGURE.FILE MINIMUM.MODULUS ... (in the Prime/PI Open version) would accept a keyword of IMMEDIATE to force the splitting of groups. Universe lacks this option, so you should specify MINIMUM.MODULUS at the time that you create the part file. --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ --- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
