On Fri, Jan 15, 2010 at 4:22 PM, Rich Rumble <[email protected]> wrote: > I upgraded the server to 2.3 Dec 10th 2009. The 10 new clients are > running 2.3 and seem unaffected, all others are 2.2 clients(windows). > Again they are logging their hash changes etc to the server while they > are connected, but they are cycling through being connected and not > connected (no network issues for the server or clients, up time is > 100% from a network and OS level.) I'm not sure if nagios is checking > inodes, can 50gig's (out of 100G) of inodes have something happen to > them? The server is running Ext3 if it matters raid10, I don't know > this is a bit out my depth now. It's just all so coincidental to have > happen all at midnight new years, stranger things have happened I > suppose. > I'll try running in debug, and I'll try to add a new 2.2 test client > and see if the issue is truly isolated to 2.2 clients, or simply to > clients that were connected before 1/1/2010 for us. >
I've had systems with plenty of free gigabytes on a storage volume run out of inodes. It can be an issue on partitions that contain a LOT of small files. It may not have been an issue this time. Might be worth keeping an eye on though depending on the system's workload.
