OpenVZ Forum: Support » nfs caching/delay from within a container

Home » General » Support » nfs caching/delay from within a container

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

nfs caching/delay from within a container [message #35320]

Tue, 17 March 2009 13:47

kclair
Messages: 2
Registered: March 2009

Junior Member

Hello,

My kernel version is 2.6.18-8.1.8.el5.028stab039.1.

We use openvz containers in a shared hosting environment. The containers mount the same nfs filesystems that our other non-openvz servers use. In particular, our ftp server runs on a different machine than the openvz servers, and our customers who are in containers use the ftp server to upload files, while their sites are served from the container servers.

We've had intermittent reports of files being uploaded via ftp that have a delay before they are seen from within the container. We've also had reports of new versions of files being uploaded with ftp, and the new versions are seen by downloading the files via ftp while the old versions are seen when downloaded from a browser (from the website within the container).

I've been able to verify that the customer reports are accurate at least several times. However, as I said, it does seem intermittent.

I'm just wondering if anyone else has noticed anything like this.

Thanks,
Kristina

Report message to a moderator

Re: nfs caching/delay from within a container [message #35352 is a reply to message #35320]

Wed, 18 March 2009 17:27

maratrus
Messages: 1495
Registered: August 2007
Location: Moscow

Senior Member

I've never heard about this but the kernel you're using is quite old. If you have a chance check the newest one.

Report message to a moderator

Re: nfs caching/delay from within a container [message #36038 is a reply to message #35320]

Wed, 13 May 2009 15:03

kclair
Messages: 2
Registered: March 2009

Junior Member

we updated our kernel and are now using 2.6.18-92.1.18.el5.028stab060.8ent. however, the problem persists.

i do notice a lot of rpc.statd and lockd errors, such as the following:

rpc.statd[6462]: STAT_FAIL to www-vps for SM_MON of 192.168.0.4
kernel: lockd: cannot monitor 192.168.0.4

when this occurs, the problem is seen both from within the vps and from the openvz host server (the nfs filesystems are mounted with --bind in the vps). oddly, doing an ls -l of the directory where the problem is happening seems to jolt the server into reality, "fixing" the problem.

i definitely acknowledge that this could be a problem unrelated to openvz, but our openvz servers are our only servers with this problem.

is anyone else using nfs with openvz in a high-use environment where the main data (in this case website files) is stored on nfs?

thanks!

Report message to a moderator

Re: nfs caching/delay from within a container [message #36045 is a reply to message #36038]

Thu, 14 May 2009 13:24

abulava
Messages: 3
Registered: May 2009

Junior Member

kclair wrote on Wed, 13 May 2009 18:03

the nfs filesystems are mounted with --bind in the vps

Do you mean that NFS mount points aren't mounted inside VE natively? I have a following setup: a web server in VE mounts a R/W NFS share from another physical server according to http://wiki.openvz.org/NFS#NFS_client

Works for me.

Report message to a moderator

Previous Topic:	port forwarding (http servers on each VPS)
Next Topic:	Error in Quick Installation instructions?

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Fri Jul 25 15:08:31 GMT 2025

Total time taken to generate the page: 0.34715 seconds