<div dir="ltr">Quite strangely, I found 2 good hosts (successfully mount the file system), where the TCP ping goes through on one, while it doe snot on the other (though LNET ping is OK for both).<div><br></div><div>- Youssef</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Jun 21, 2023 at 6:08 PM Youssef Eldakar <<a href="mailto:youssefeldakar@gmail.com">youssefeldakar@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div dir="ltr">Thanks, Rick, for that suggestion. TCP ping between a problematic host and the MDS indeed does not go through.</div><div dir="ltr"><br></div><div>Not exactly sure what to investigate next, but that gives me somewhere to start...</div><div><br></div><div>- Youssef</div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Jun 20, 2023 at 7:00 PM Mohr, Rick via lustre-discuss <<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Have you tried tcp pings on the IP addresses associated with the IB interfaces?<br>
<br>
--Rick<br>
<br>
<br>
On 6/20/23, 12:11 PM, "lustre-discuss on behalf of Youssef Eldakar via lustre-discuss" <<a href="mailto:lustre-discuss-bounces@lists.lustre.org" target="_blank">lustre-discuss-bounces@lists.lustre.org</a> <mailto:<a href="mailto:lustre-discuss-bounces@lists.lustre.org" target="_blank">lustre-discuss-bounces@lists.lustre.org</a>> on behalf of <a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a> <mailto:<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a>>> wrote:<br>
<br>
<br>
In a cluster having ~100 Lustre clients (compute nodes) connected together with the MDS and OSS over Intel True Scale InfiniBand (discontinued product), we started seeing certain nodes failing to mount the Lustre file system and giving I/O error on LNET (lctl) ping even though an ibping test to the MDS gives no errors. We tried rebooting the problematic nodes and even fresh-installing the OS and Lustre client, which did not help. However, rebooting the MDS seems to possibly momentarily help after the MDS starts up again, but the same set of problematic nodes seem to always eventually revert back to the state where they fail to ping the MDS over LNET.<br>
<br>
<br>
Thank you for any pointers we may pursue.<br>
<br>
<br>
<br>
<br>
Youssef Eldakar<br>
Bibliotheca Alexandrina<br>
<a href="http://www.bibalex.org" rel="noreferrer" target="_blank">www.bibalex.org</a> <<a href="https://urldefense.us/v2/url?u=http-3A__www.bibalex.org&amp;d=DwMFaQ&amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;s=5DLPIzJx0tgg1TgSZkvvNNVfDfgpo-Prv-BPOga0WMA&amp;e=" rel="noreferrer" target="_blank">https://urldefense.us/v2/url?u=http-3A__www.bibalex.org&amp;d=DwMFaQ&amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;s=5DLPIzJx0tgg1TgSZkvvNNVfDfgpo-Prv-BPOga0WMA&amp;e=</a>> <<a href="https://urldefense.us/v2/url?u=http-3A__www.bibalex.org&amp;amp;d=DwMFaQ&amp;amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;amp;s=5DLPIzJx0tgg1TgSZkvvNNVfDfgpo-Prv-BPOga0WMA&amp;amp;e=&gt" rel="noreferrer" target="_blank">https://urldefense.us/v2/url?u=http-3A__www.bibalex.org&amp;amp;d=DwMFaQ&amp;amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;amp;s=5DLPIzJx0tgg1TgSZkvvNNVfDfgpo-Prv-BPOga0WMA&amp;amp;e=&gt</a>;><br>
<a href="http://hpc.bibalex.org" rel="noreferrer" target="_blank">hpc.bibalex.org</a> <<a href="https://urldefense.us/v2/url?u=http-3A__hpc.bibalex.org&amp;d=DwMFaQ&amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;s=HMqKriFlJ2qwafMOSVJMqre9-wmJ--kaSS_rx4t7hQw&amp;e=" rel="noreferrer" target="_blank">https://urldefense.us/v2/url?u=http-3A__hpc.bibalex.org&amp;d=DwMFaQ&amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;s=HMqKriFlJ2qwafMOSVJMqre9-wmJ--kaSS_rx4t7hQw&amp;e=</a>> <<a href="https://urldefense.us/v2/url?u=http-3A__hpc.bibalex.org&amp;amp;d=DwMFaQ&amp;amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;amp;s=HMqKriFlJ2qwafMOSVJMqre9-wmJ--kaSS_rx4t7hQw&amp;amp;e=&gt" rel="noreferrer" target="_blank">https://urldefense.us/v2/url?u=http-3A__hpc.bibalex.org&amp;amp;d=DwMFaQ&amp;amp;c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&amp;amp;r=SpEwA4Pnyq7nH7aMGq8KpA&amp;amp;m=kwZRPirpHWOowgLmVOYe_KJ4ZigAHQk3DiF8-BwQ2qFikINn8C5-0SyyYEDelqDH&amp;amp;s=HMqKriFlJ2qwafMOSVJMqre9-wmJ--kaSS_rx4t7hQw&amp;amp;e=&gt</a>;><br>
<br>
<br>
<br>
<br>
<br>
_______________________________________________<br>
lustre-discuss mailing list<br>
<a href="mailto:lustre-discuss@lists.lustre.org" target="_blank">lustre-discuss@lists.lustre.org</a><br>
<a href="http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org" rel="noreferrer" target="_blank">http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org</a><br>
</blockquote></div></div>
</blockquote></div>