Re: [Veritas-vx] inode invalid?!
Just as an FYI my preferred order for fscks # fsck -F vxfs /dev/vx/rdsk/DG_name/VOL_name Intent LOG ONLY replay # fsck -F vxfs -o full,nolog -n /dev/vx/rdsk/DG_name/VOL_name FILE SYSTEM AND METADATA STRUCTURES ONLY - NO INTENT LOG - a LOOK at what will happen remembering there ARE transactions in the intent log. # fsck -F vxfs -o full -n /dev/vx/rdsk/DG_name/VOL_name ALL FILE SYSTEM AND METADATA STRUCTURES - a LOOK at what will happen You can mount READ ONLY for a quick look - or to backup crucial data - remember unplayed transaction in the intent log # mount -F vxfs -o ro /dev/vx/dsk/DG_name/VOL_name /mnt You can ff / ncheck to match file names to inodes removed in above fsck -o full # ncheck -F vxfs /dev/vx/dsk/DG_name/VOL_name # ff -F vxfs /dev/vx/dsk/DG_name/VOL_name This is usefult to cross reference file that will be removed in the full fsck - these are in the fsck output above. Check what is in the intent log # echo fmtlog | fsdb -F vxfs /dev/vx/rdsk/DG_name/VOL_name Remember fset 1 is the VxFS metadata, fset 999 is user data. Collect a METASAVE - use the binary in /opt/VRTSspt/FS/MetaSave # fsck -F vxfs -o full,nolog -y /dev/vx/rdsk/DG_name/VOL_name # fsck -F vxfs /dev/vx/rdsk/DG_name/VOL_name ALL FILE SYSTEM AND METADATA STRUCTURES - WITH REMOVING STUFF IF THERE ARE ERRORS OR # fsck -F vxfs -o full -y /dev/vx/rdsk/DG_name/VOL_name ALL FILE SYSTEM AND METADATA STRUCTURES - WITH REMOVING STUFF IF THERE ARE ERRORS FMTLOG - Internal Technote 251996 -Original Message- From: veritas-vx-boun...@mailman.eng.auburn.edu [mailto:veritas-vx-boun...@mailman.eng.auburn.edu] On Behalf Of Thomas Graham Sent: Wednesday, 3 March 2010 4:25 PM To: William Havey Cc: veritas-vx@mailman.eng.auburn.edu Subject: Re: [Veritas-vx] inode invalid?! I see, so checking lost and found dir is an important step On Tuesday, March 2, 2010, William Havey bbha...@gmail.com wrote: The purpose of fsck is to make a file system mountable. If this requires putting some unaccounted for file system block into the lost+found directory, leaving the admin to decide what to do with the block, so be it. A file system simply can not guarantee data integrity under every conceivable circumstance. That's why redundancy is built into either the hardware, if the situations cause is hardware failure, the array provides a good copy, or the storage software, through mirroring. On Tue, Mar 2, 2010 at 7:18 AM, Thomas Graham lktho...@gmail.com wrote: But would it remove some blocks which contain data? On Tuesday, March 2, 2010, William Havey bbha...@gmail.com wrote: A file system check of VxFS should be very quick. Read the Intent Log records, write them out to disk, done. Do the backup after fsck. On Sun, Feb 28, 2010 at 9:31 PM, Thomas Graham lktho...@gmail.com wrote: guys, I am using veritas cluster file system, and currently CFS is marked as dirty and I have to take the whole FS offline to do fsck scanning. Anyone know if there have any risk to hit inode if I do backup during inode invalid situation? -- Thomas G Lau Tel: 93239670 ___ Veritas-vx maillist - veritas...@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx -- Thomas G Lau Tel: 93239670 -- Thomas G Lau Tel: 93239670 ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx
Re: [Veritas-vx] Migrating volumes on windows servers [SEC=UNCLASSIFIED]
Greg I am not sure that is ever going to work - but there are some things to look at. In CDS format (Unix) the private region is at the beginning of disk on the whole disk partition ( 65536 blocks in VxVM Ver 5.x, 2048 blocks in VxVM 4.x ). # vxdisk list c3t50060E80004372C0d25s2 public:slice=2 offset=65792 len=855808 disk_offset=0 private: slice=2 offset=256 len=65536 disk_offset=0 Note - in above the 65536 blocks has 256 blocks offset - for AIX headers if CDS sharing between these two platforms but at disk_offset=0. My LUN geometry is 512 bytes/sector, 768 sectors/track, 50 tracks/cylinder, 38400 sectors/cylinder. So the CDS initialization does NOT necessarily align to cylinder boundary, and remember this is CDS initialised - not sliced format. Windows may need this - so this may be a pre-requisite for the attempted migration. There are some exclusions for reserved region for other platforms - AIX headers and others but AFAIK these are now offset to the existing private region as shown above. I would have to check on the layout etc for a VxVM under Windows and what size / location / offset it uses and in basic / dynamic. There does seem to be some problems with what you are proposing here - namely the partition requirement for dynamic - remembering also that CDS on Linux and Solaris uses a Solaris partition table. In Unix the objects defined in that diskgroup - these are just defined types and offset / lengths on the actual disk - so a volume has a subdisk object specifying the storage LUN ID and offset and length in the public region, a plex object to define the volume redundancy and a volume object to define an IO object to the controlling OS ( major number pointing back to the vxio driver and specific minor for the specific volume number from the group minimum minor - some caveats here for shared diskgroups etc ) . If the same LUN or a mirror is then presented to SF Windows ( which AFAIK there is no direct migration ) but if the initialization on Win server once LUNs are presented can be made to fill the same size as Unix private region - then remaking the volumes on Win should just remake the offsets and lengths of the objects and appropriate layers. So - keep initialization area same size / offset. Remake objects and make sure to not use any init=zero on either side. Just some things to keep in mind if attempting this. Stuart -Original Message- From: veritas-vx-boun...@mailman.eng.auburn.edu [mailto:veritas-vx-boun...@mailman.eng.auburn.edu] On Behalf Of Robinson, Greg Sent: Tuesday, 16 March 2010 4:24 PM To: veritas-vx@mailman.eng.auburn.edu Subject: [Veritas-vx] Migrating volumes on windows servers [SEC=UNCLASSIFIED] UNCLASSIFIED Hi all, I was migrating a volume from one storage array to another and from one server to another, under windows vxvm, and it didn't work. I followed my normal procedure which I use for my UNIX servers, but windows vxvm did something different that I was not expecting. I mirrored the data in the normal way, broke the mirror, removed the disks from the disk group (and this is where I think windows vxvm got the better of me), and tried to import the disks and create a new disk group out of them and mount my volume. As soon as I removed the disks from the disk group, windows vxvm converted them into basic disks instead of leaving them as dynamic disks. My question is: how can I migrate the volume without deporting the disk group from one server and importing it from another. The trouble is, we don't have access to the other storage array, but we do the server. And access control on the other array prevents us from importing all the disks in question. Also, I should note that VCS is running on both servers, but that was not a problem. Thankx, Greg. IMPORTANT: This email remains the property of the Australian Defence Organisation and is subject to the jurisdiction of section 70 of the Crimes Act 1914. If you have received this email in error, you are requested to contact the sender and delete the email. ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx
Re: [Veritas-vx] VxVm
Colin OK - I see you are using enclosure naming - and BTW - the /dev/vx/dmp /dev/vx/dmp are just the same as disks BUT with the added provision of DMP to keep the device online. How many paths to these devices # vxdmpadm getsubpaths or # vxdisk path If a CLARiiON , and if only 2 paths - then set the array iopolicy to singleactive - this is most likely - see later for DMP # vxdmpadm listenclosure all # vxdmpadm setattr enclosure ENC_name iopolicy=singleactive If a CLARiiON , and if more than 2 paths - then set the array iopolicy to balanced - and DMP does know how to stop IO to the secondary paths. Note - this is against EMC recommendations but it works. # vxdmpadm listenclosure all # vxdmpadm setattr enclosure ENC_name iopolicy=balanced I also notice that the enclosure names are lower case - indicating a V5.x release VxVM installed. - are the CLARiiON APMs running ? # vxdmpadm listapm all Check the CLARiiON are Active If a fencing cluster - then these are local LUNs at the point of /dev/vx/dmp names for the devices. And yes SCSI3 keys will be placed on them in a fencing cluster # gabconfig -a Check if there is Port b membership - if so then yes you have a fencing cluster. Check also the DMP block switch - it may be that with the iopolicy incorrect, and on low IO you did not reach the limit for # vxdmpadm gettune all dmp_pathswitch_blks_shift Now - when busy and IO chunks bigger than path switch level ( and with iopolicy incorrect ) a path switch will cause a trespass (check SAN logs) AND a block, drain, resume on the DMP path. There will be a failover message logged in /etc/vx/dmpevents.log - check here also. Stuart From: veritas-vx-boun...@mailman.eng.auburn.edu [mailto:veritas-vx-boun...@mailman.eng.auburn.edu] On Behalf Of Collin Sent: Wednesday, 31 March 2010 2:09 AM To: William Havey Cc: veritas-vx@mailman.eng.auburn.edu Subject: Re: [Veritas-vx] VxVm Sorry for any confusion... I've got several powerpath devices from a dead system that I'm mounting temporarily on one node in my cluster. I run a devfsadm -Cv and vxdctl enable. After that I can see the powerpath devices listed as... emc_clarrion0_10 auto:none - -online invalid I modified my /etc/vfstab file to mount the devices.. /dev/vx/dmp/emc_clariion0_10s6 /dev/vx/rdmp/emc_clariion0_10s6 /u10 ufs 3 yes - The device mounts and I can access the file system with all my data. When the activity starts to increase on these temporary mount points, I see a count down on the console that port H has lost connectivity. After the 16 seconds, the node panics and of course reboots. However, if I mount the power path devices using a single path.. /dev/dsk/c1t5006016100600432d10s6 /dev/rdsk/c1t5006016100600432d10s6 /u10 ufs 3 yes - I never get the port H losing connectivity. I want to use the dmp name in case I lose a path to these disk. Any reason why using the dmp name causes port H to lose connectivity vs. using a single path? Thanks, Collin On Tue, Mar 30, 2010 at 10:48 AM, William Havey bbha...@gmail.com wrote: The original message states mount these disks as /dev/vx/dmp/emc_array_Xs6 . Perhaps this is normal behavior. Mounts are of devices which receive I/O. A /dev/vx/dmp/... device entry isn't I/O capable. I think a clearer statement of what Collin intends to do is needed. Bill On Tue, Mar 30, 2010 at 3:01 AM, Dmitry Glushenok gl...@jet.msk.su wrote: Hello, Panic string and previous messages usually helps to understand cause.. Release notes to RP2-RP3 also provides short descriptions of fixed issues like Fixed the cause of a system panic when mutex_panic() was called from vol_rwsleep_wrlock(). On 29.03.2010, at 19:02, Collin wrote: I've got the following Solaris 10 VxVM 5.0MP3RP1HF12 I have a number of mount points that are being migrated from /dev/dsk/cXtXdXsX to clustered mount points. The problem I'm having is if I mount these disk in the /dev/dsk/cXtXdXsX format I run the risk that if something were to cause the direct path to go down I would lose the databases on these mount points. But when I mount these disks as /dev/vx/dmp/emc_array_Xs6 my system panics and core dumps. Does VxVM have any issues mounting /dev/vx/dmp/emc_array_Xs6?? Thanks, Collin ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx -- Dmitry Glushenok Jet Infosystems ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx
Re: [Veritas-vx] Linux VxVM Setup
Martin Some things to look at - are the read_ahead VxFS tuning - this is optimised to sequential reads. Other VxFS tunables are here http://support.veritas.com/docs/344352 The usual one is max_direct_iosz for performance. Also you could vxtrace the volume plex subdisk operations while the backups are going on and / or vxdmpadm iostat the LUNs / paths in particular and also iostat -Cxn to see which layer the slow down is occurring at. Stuart From: veritas-vx-boun...@mailman.eng.auburn.edu [mailto:veritas-vx-boun...@mailman.eng.auburn.edu] On Behalf Of Martin, Jonathan Sent: Saturday, 29 May 2010 12:01 AM To: veritas-vx@mailman.eng.auburn.edu Subject: Re: [Veritas-vx] Linux VxVM Setup Thanks for the help. Your post and another emailed to me privately led me to find an issue with the multipath driver. I didn't notice this before, but there should not be an /dev/sdc. This was an alternate path to the same lun. I've got my volume configured, 1.5 TB of data restored and flashbackups are running. However, the speed isn't what I had hoped. The backup runs for about an hour at 1MB/sec, backing up what looks like metadata. After that, the speed ups to ~25MB/sec. If I backup the data directly off the lun with a standard policy I get a steady 30MB/sec. Are there any configuration settings or logs I can look through? Are there buffer settings I can toy with? There does not seem to be much in /etc/vx/log. We've had quite a bit of success in similar flashbackup scenarios with Windows file servers, and we're hoping to push past 30MB/sec on Linux too. Thanks, -Jonathan From: William Havey [mailto:bbha...@gmail.com] Sent: Wednesday, May 26, 2010 3:08 PM To: Martin, Jonathan Cc: veritas-vx@mailman.eng.auburn.edu Subject: Re: [Veritas-vx] Linux VxVM Setup Jonathan, Use fdisk to clear up the error state then initialize the disks. Bill On Wed, May 26, 2010 at 10:51 AM, Martin, Jonathan jmart...@intersil.com wrote: Greetings all, first time poster here, so please be gentle. I'm trying to run a POC for NetBackup Linux Flashbackup, but to do that I need a VxVM volume and VxFS partition. I've got a 6TB lun presented to a test RedHat 2.6 server as /dev/sdb. When I run vxdiskadm, option 1 to initialize the disk I get the following error. This disk device does not appear to be valid. The disk may not have a valid or usuable partition table, the special device file for the disk may be missing or invalid, or the device may be turned-off or detached from the system. This disk will be ignored. Output format: [Device_Name,Disk_Access_Name] [sdb,sdb] vxdisk list gives me the following output. DEVICE TYPEDISK GROUPSTATUS sda auto:none --online invalid sdb auto--error sdc auto--error I also tried running through the VxVM Admin guide and got the following: vxdisk init sdb VxVM vxdisk ERROR V-5-1-0 read of lvm header blocks for /dev/vx/rdmp/sdb failed VxVM vxdisk ERROR V-5-1-5433 Device sdb: init failed: Disk Ioctl failed My Symantec rep gave me some trial keys and free software, but I'm on my own for configuration. Can someone here throw me a bone? I've got to be doing something wrong. Thanks! -Jonathan ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx
Re: [Veritas-vx] vxrepquota question
It works #umount /mnt #mount -F vxfs /dev/vx/dsk/testdg/vols /mnt #vxquot /mnt /dev/vx/rdsk/testdg/vols: USERS 11266 root 4096 adm 4096 bin 4096 cstynes 4096 gdm 4096 daemon 4096 joe 4096 listen 4096 lp 4096 noaccess 4096 nobody 4096 nobody4 4096 nuucp 4096 oracle 4096 postgres 4096 smmsp 4096 svctag 4096 sys 4096 uucp 4096 webservd 3072 Admin Just a quick test after remounting - it works Generated a small amount of junk using the following for (( i=0; i = 512; i++ )) ; do echo $i; cat /etc/passwd | cut -d : -f1 | while read user; do dd if=/dev/zero of=$user.$i bs=1024k count=1; chown $user $user.$i; done; done From: veritas-vx-boun...@mailman.eng.auburn.edu [mailto:veritas-vx-boun...@mailman.eng.auburn.edu] On Behalf Of John Cronin Sent: Saturday, 5 March 2011 10:16 AM To: Carl E. Ma Cc: veritas-vx@mailman.eng.auburn.edu Subject: Re: [Veritas-vx] vxrepquota question Try the command vxquot file-system. According to the man page, it does not appear that quotas need to be enabled for this command to work. Unfortunately, I don't have any place to try this out right now. On Fri, Mar 4, 2011 at 4:40 PM, Carl E. Ma zhu_ju...@yahoo.ca wrote: Hello All, We are running NFS server with Veritas VCS 5.1SP1 on solaris 10 x86. In order to track disk usage in real time, we enabled disk quota on the server so that we can get each user's usage with vxrepquota filesystem and vxquota -v -u username. If we didn't set quota for a user, his name/usage won't show up in vxrepquota output. Since we don't know how many users will keep files on the shared NFS filesystem, we have to enable quota for all 3000+ users as temp solution. My questions is without enforcing soft/hard quota, can we still track user disk usage? My understanding of quota filesystem is all users' disk usage is being tracked within filesystem and there should have other way to read out the statistics. I will summarize if there is answer. Thanks have a good weekend, zhu ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx ___ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx