Whamcloud Community JIRA

Whamcloud Community JIRA https://jira.whamcloud.com This file is an XML representation of an issue en-us

9.4.14

940014

05-12-2023

[LU-1673] Locking issue with 1.8.x clients talking to 2.2 Servers https://jira.whamcloud.com/browse/LU-1673 Lustre <p>We noticed that clients running Lustre 1.8.x seem to have trouble locking files hosted on 2.2 Servers.</p> client 1.8.x server 2.2 LU-1673

Locking issue with 1.8.x clients talking to 2.2 Servers

Bug Critical Resolved Fixed Yang Sheng ETHz Support client locking server Wed, 25 Jul 2012 12:37:38 +0000 Mon, 19 Nov 2012 11:39:00 +0000 Mon, 19 Nov 2012 11:39:00 +0000 Lustre 2.2.0 0 3 <p>This simple C code is enough to reproduce the problem:</p> <p>$ cat test.c <br/> #include <sys/file.h><br/> #include <errno.h></p> <p>int main() {<br/> int fd;<br/> int r;</p> <p> printf("-- starting --\n");</p> <p> fd = open("locktest.txt", O_RDWR);<br/> if(fd < 0) </p> { printf("open failed\n"); exit(1); } <p> r = flock(fd, LOCK_EX|LOCK_NB);<br/> if(r == -1) </p> { printf("Error in flock: %d\n", errno); exit(1); } <p> flock(fd, LOCK_UN);<br/> close(fd);<br/> return 0;<br/> }</p> <p>Creating 'locktest.txt' on a 2.2 server (while using the 1.8) client + starting the application ~2-3 times causes flock() to fail:</p> <p>rm -f locktest.txt ; touch locktest.txt ; for x in </p> {0..5} <p> ; do ./a.out ; sleep 1 ; done<br/> – starting –<br/> -- starting –<br/> Error in flock: 11<br/> – starting –<br/> Error in flock: 11<br/> – starting –<br/> Error in flock: 11<br/> – starting –<br/> Error in flock: 11<br/> – starting –<br/> Error in flock: 11</p> <p>The 'EAGAIN' error will be gone after a couple of seconds (i suppose that's when the leaked<img class="emoticon" src="https://jira.whamcloud.com/images/icons/emoticons/help_16.png" height="16" width="16" align="absmiddle" alt="" border="0"/> lock timed out).</p> <p>Note that exactly the same code works fine on:</p> <ul class="alternate" type="square"> <li>A 1.8.x client talking to 1.8.x servers</li> <li>A 2.2 client talking to 2.2 servers</li> </ul> <p>The 1.8 client in my test is running:<br/> $ cat /proc/fs/lustre/version <br/> lustre: 1.8.7.80<br/> kernel: patchless_client<br/> build: ../lustre/scripts--PRISTINE-2.6.18-308.1.1.el5</p> <p>The 2.2 servers are on:</p> <p>bash-4.1$ uname -r<br/> 2.6.32-220.4.2.el6_lustre.x86_64</p> <p>bash-4.1$ cat /proc/fs/lustre/version <br/> lustre: 2.2.0<br/> kernel: patchless_client<br/> build: 2.2.0-RC2--PRISTINE-2.6.32-220.4.2.el6_lustre.x86_64</p> <p>The filesystem is mounted via:<br/> $ grep _xl /etc/fstab<br/> 10.201.62.13@o2ib:10.201.62.14@o2ib:/nero /cluster/scratch_xl lustre flock,_netdev 0 0</p> <p>Oleg is look into this one</p> <p>Hm, I was under impression that the fix for this landed in time for 2.2, but alas.</p> <p>The patch that fixes this could be found here: <a href="http://review.whamcloud.com/#change,2193" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#change,2193</a></p> <p>Would we have to patch only the MDS or also all OSTs?<br/> Upgrading all OSTs requires quite some downtime :-/</p> <p>Could you provide us the rpms patched for 2.2 servers? </p> <p>Thanks in advance.</p> <p>flocks are only taken on MDS, so updating just MDS is fine.</p> <p>Could you provide us patched rpms?</p> <p>Thanks in advance</p> <p>Yangsheng is working on creating patched RPMs</p> <p>Don't forget the patch for <a href="http://review.whamcloud.com/#change,3008" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#change,3008</a> as well since it is need to 2.3 <-> 2.2 interop testing,</p> <p><a href="http://review.whamcloud.com/#change,3486" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#change,3486</a></p> <p>James, like you correctly mention, that change is only needed for 2.3 interop which is not the case here, so there is no rush to get it included, esp. since there is no official 2.3 release and won't be for some time.</p> <p>You can get RPMs here: <a href="http://build.whamcloud.com/job/lustre-reviews/7975/arch=x86_64,build_type=server,distro=el6,ib_stack=inkernel/artifact/artifacts/RPMS/" class="external-link" target="_blank" rel="nofollow noopener">http://build.whamcloud.com/job/lustre-reviews/7975/arch=x86_64,build_type=server,distro=el6,ib_stack=inkernel/artifact/artifacts/RPMS/</a></p> <p>As Lurii commented in gerrit:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>Iurii Golovach Nov 2 Patch Set 1: I would prefer that you didn't submit this This patch looks obsolete since there are already number of patches which cover this issue: http://review.whamcloud.com/#change,3722 http://review.whamcloud.com/#change,3202 http://review.whamcloud.com/#change,3203 http://review.whamcloud.com/#change,3725 http://review.whamcloud.com/#change,3727 </pre> </div></div> <p>So close this bug.</p> Development Epic client locking server Rank 1|hzv34v: Rank (Obsolete) 4007 Severity