https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-11382 Lustre <p>In a cluster with more than one router what is the most graceful way to reboot some or all of the routers, without causing client/server connective issues.</p> <p> </p> <p> </p> LU-11382

Graceful Router Reboots

Question/Request Minor Resolved Fixed Sonia Sharma Mahmoud Hanafi Fri, 14 Sep 2018 21:21:50 +0000 Wed, 24 Jul 2019 17:13:18 +0000 Wed, 24 Jul 2019 17:13:18 +0000 0 4 <p>Sonia</p> <p>Could you please advise?</p> <p>Thanks</p> <p>Peter</p> <p>HI Mahmoud,</p> <p> To reboot a router without affecting the client-server connectivity, ideally, we would need to ensure that an alternate routing path is available for client-server connectivity. And it would be good to bring down the routes configured with the router (remove the related routes configured on clients and servers) before rebooting that specific router. If there are multiple routers that need to be rebooted, it would be better to reboot them one by one instead of rebooting all at once.</p> <p> Apart from this general advice, it would help to know the what the network configuration of the system looks like. There was a bug related to issues resulting after router reboots which got fixed in lustre 2.8. So it would also depend on the lustre version running on the nodes that whether or not a router reboot result in client-server connectivity issues.</p> <p> A bug that I was referring to is - <br/> <a href="https://jira.whamcloud.com/browse/LU-7646" class="external-link" rel="nofollow">https://jira.whamcloud.com/browse/LU-7646</a> - This bug is seen with mlx5 cards wherein when a router is rebooted, it is possible for a client's attempt to reconnect to the router to get stuck in a permanent connecting state. When the router comes up and tries to create a connection back to the client, that connection will be rejected as CON RACE. And this goes in an infinite loop.</p> <p>please close.</p> Development Rank 1|i002if: Rank (Obsolete) 9223372036854775807