Summary: | Slurm 18.08.0 - X11 connection rejected because of wrong authentication. | ||
---|---|---|---|
Product: | Slurm | Reporter: | Lee Hobson <lhobson> |
Component: | Scheduling | Assignee: | Jacob Jenson <jacob> |
Status: | RESOLVED INFOGIVEN | QA Contact: | |
Severity: | 6 - No support contract | ||
Priority: | --- | CC: | luca.capello, support |
Version: | 18.08.0 | ||
Hardware: | Linux | ||
OS: | Linux | ||
Site: | -Other- | Alineos Sites: | --- |
Atos/Eviden Sites: | --- | Confidential Site: | --- |
Coreweave sites: | --- | Cray Sites: | --- |
DS9 clusters: | --- | HPCnow Sites: | --- |
HPE Sites: | --- | IBM Sites: | --- |
NOAA SIte: | --- | OCF Sites: | --- |
Recursion Pharma Sites: | --- | SFW Sites: | --- |
SNIC sites: | --- | Linux Distro: | --- |
Machine Name: | CLE Version: | ||
Version Fixed: | Target Release: | --- | |
DevPrio: | --- | Emory-Cloud Sites: | --- |
Description
Lee Hobson
2018-09-13 08:22:48 MDT
Just a quick follow up comment on this. Taken from https://groups.google.com/forum/#!topic/slurm-users/cpzsGqoqcCI for reference - bottom post by Marcus Wagner: If I run the following on a forwarded SSH session (not x2Go) the job does start. salloc srun --x11 'xterm' Is there any reasoning behind this, or is this the intended way X11 support in Slurm should work? Interestingly, if I run the same command on an X2Go SSH remote desktop session, I will then see the following: salloc: Pending job allocation 200145 salloc: job 200145 queued and waiting for resources salloc: job 200145 has been allocated resources salloc: Granted job allocation 200145 salloc: Waiting for resource configuration salloc: Nodes rjm-compute001 are ready for job srun: error: Cannot forward to local display. Can only use X11 forwarding with network displays. salloc: Relinquishing job allocation 200145 Hi Lee, Do you work with Chris Hardacre at OCF? We noticed that you did not tag a site when you opened this issue and this has put this ticket into an unsupported status. We are trying to confirm support for this issue and your response will help us expedite that process. -Jason Hi, Yes I do, apologies for not updating that properly. Thanks, Lee (lhobson@ocf.co.uk) Lee, Can you please verify which Site this request is from? Thanks, Jacob I assume you are referring to which OCF site, this request has come from the Sheffield office. I'm currently on annual leave until the 29th October 2018 with very limiited access to email. Please contact Chris Devine/Faye Exton (project manager), Chris Hardacre (Support) or Russel Slack (Operations Director) should your query be urgent. Hi there, at the University of Geneva (Switzerland) we are experiencing the very same issue with 18.08.1, but without the --x11 option. I have just seen that the latest stable version is 18.08.3, I will build that and come back with the results. Thx, bye, Luca Hi there, (In reply to Luca Capello from comment #7) > I have just seen that the latest stable version is 18.08.3, I will build > that and come back with the results. No changes, X11 forwarding is disabled without the --x11 option and with I get the same error message as in comment #0 : ===== srun: error: Cannot forward to local display. Can only use X11 forwarding with network displays. ===== Thx, bye, Luca Hi Luca and Lee, There are a number of changes coming in 19.05 that address issues like this with X11. Here are some of the commits related to the re-work. 9c8be2689e078756d020d19d8fb9ab2c09a88be5 91170a04641d28d8020d1e4708af080ceb1e3279 f2da4d7c174a0baf4e15301b947e5625fb747c56 c97284691b6a0df57493a13132787a1a908a749f 2a58e3e228c4b0b589e2d6456159fe725e21d32d 3b7d1625c470d479d1c5d8cb492ae8918d551d7f 6985ccbac42a442c73fe91d5ee6146fe901058f1 We have tracked this via Bug #3647 as well. In regards to the error message: > srun: error: Cannot forward to local display. Can only use X11 forwarding with network displays. Our current X11 forwarding implementation cannot connect to unix sockets at this time. Two options if you plan to continue using this version are: - Use "ssh -X localhost", then run "srun --x11" within that SSH session. SSH itself will handle translation between a TCP socket that Slurm's implementation can use to the local unix socket. - Disable our build-in integration, and use the SPANK X11 plugin instead. Due to differences in how it forwards traffic, it can accommodate use of a unix socket instead of a network socket. -Jason |