Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cross-project-issues-dev] Hudson home on a slow NFS mount

On 02/24/2012 08:39 AM, Denis Roy wrote:
On 02/23/2012 04:08 PM, Winston Prakash wrote:
I see that the Hudson jobs folder is at

 jobs -> /opt/public/jobs

which seems to be a NFS mount from

wilma:/opt/public    1975741280 1248514496 626864960  67% /opt/public
This NFS mount seems to be really slow.

The NFS mount itself is fine -- it's the underlying hardware that is "slow". Although /opt/public is an entreprise-class 16-disk SCSI RAID, it is from 2004, so by today's standards it's quite slow.

I may have a no-cost, perfectly transparent solution, though... That NFS server has a 3-disk 250G RAID array which is perfectly idle.

The "idle" array produces your command output in 9 minutes, compared to 40+ on /shared (which is busy). This is not spectacular, but it is a magnitude faster than its current location.

Your find command must walk the tree of 2,500,000 directories and files and stat() each one. This is not typical of a build job, so although moving Hudson's jobs/ directory to this array will definitely help, I'm not convinced it will be noticeable.

But like I always say, every bit helps. We'll do the move towards the end of the day/weekend. From Hudson's point of view, the new array will be mounted in the same location, so nothing changes.

By way of comparison, the same command takes 3 minutes on our primary NFS server, but I don't want to host build data here, by fear that a busy Hudson would impact performance of CVS, Git, Bugzilla and everything else.


Back to the top