We have 20 Linux clients mounting 7 NFS servers(3 Netapp, 4 Auspex) using the following mounting options: nfs6-e9c:/vol/vol0/qvol1 on /epdata21 type nfs (rw,proto=udp,vers=3,rsize=8192,wsize=8192,hard,intr,addr=10.200.10.140) If one of these remote NFS servers goes down, the local clients load average gets extemely high (~200) and becomes unresponsive. We have a propietery application that searches and retrieves from NFS, but we have also been able to reproduce using a simple 'cat' script. (see reproducable section) We also have Solaris clients that do the same kindof work and when an NFS server goes down the load average is unaffected unlike Red Hat Linux. Version-Release number of selected component (if applicable): kernel-smp-2.4.20-19.8 How reproducible: I will attach the simple cat script to next section. Steps to Reproduce: 1. Mount an NFS server 2. Fail NFS server(Down NIC card on server) 3. Run script Actual results: High load averages, eventaully leading to unresponesive NFS client. Expected results: Load average should not go high and should act similarly to other Unix Platforms like Solaris
Created attachment 94825 [details] Script to create high load on Linux Client This is a script that will create a high load on Linux Client after an NFS failure. "/epdata21/DO_NOT_DELETE_READ_TEST/read.test.DO.NOT.DELETE" is a flat text file on a remote NFS server file system.
Thanks for the bug report. However, Red Hat no longer maintains this version of the product. Please upgrade to the latest version and open a new bug if the problem persists. The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, and if you believe this bug is interesting to them, please report the problem in the bug tracker at: http://bugzilla.fedora.us/