Portal Home > Knowledgebase > Articles Database > Strange LVM Behavior - Slow boot
Strange LVM Behavior - Slow boot
| Posted by YUPAPA, 03-11-2014, 11:42 AM |
| I've got a large LUN - roughty 10T and it is used for storing logical volumes for backup purpose. I create a 300G logical volume one day and then we schedule the system to take a snapshot of it every single day. Over a week (e.g. Mar 2-8), I end up with 1 logical volume with 6 snapshots. The system purges the old logcial volume group and the 6 snapshots that are taken the week before (e.g. Feb 23-Mar 1).
I decide to reboot the system one day and strange things happen. During the boot phrase, it takes a long time (about 20 minutes) to for it to pass the logical volume step.
Has anyone seen this problem before? I've noticed there are archive of metadata under /etc/lvm/archive/, could those metadata together with the non-existance of volume groups (taken from Feb 23-Mar 1) lead to the lockup?
I think the only way to overcome this problem at this time is to deport the entire volume group and then reboot the system. I have not tried though. The system doesn't have many snapshots, less than 20 as it does a cleanup every week.
Any thoughts on this would be appreciated.
Last edited by YUPAPA; 03-11-2014 at 11:48 AM.
|
| Posted by YUPAPA, 03-11-2014, 11:45 AM |
| And one more thing, this occurs on both centos 5 and 6.
|
| Posted by martijnatlico, 03-11-2014, 02:46 PM |
| How many physical volumes does the volume group have? Also, are these local disks, local RAID arrays or are they connected through something like FC or iSCSI?
|
| Posted by YUPAPA, 03-12-2014, 04:13 AM |
| There is only one 10T physcial volume, underneath there is only 1 volume group and less than 20 of a mix of logical volumes and snapshots. The file operations are pretty normal, system does not freeze or lags, its just the boot process takes longer than usual.
The 10T lun is assigned from the FC-SAN. The maskings are done right and there is nothing abnormal from the SAN and switch side.
|
| Posted by martijnatlico, 03-12-2014, 11:20 AM |
| I wonder if this is caused by the snapshots, because LVM is known to have poor performance with a large number of snapshots according to a Google search.
Remember that snapshots are not backups. The most common way of using snapshots is: create snapshot, copy to backup medium/location, destroy snapshot.
|
Add to Favourites 
Print this Article
Also Read
Shared Hosting
Shared Hosting
