Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Investigate why HB is slow #122

Open
radmilak opened this issue Mar 16, 2020 · 5 comments
Open

Investigate why HB is slow #122

radmilak opened this issue Mar 16, 2020 · 5 comments
Assignees

Comments

@radmilak
Copy link

radmilak commented Mar 16, 2020

Investigate the infamous "general slowness" in HB.
What we know:

  • It does not happen constantly - seems to go in "waves" that last several hours
  • Effects all tools, though larger ProTo-tools or tools where there is a lot of computation between gui reloads are more effected due to bigger chance for ProTo to timeout (default 30s)
  • Seems at least sometimes to also effect the general HB gui (clicking in the menu, loading a tool before any input is given)
  • There might be a connection to what happens on the server - on one occasion we knew there were big jobs running when HB was very slow, pausing the jobs did not help though
  • All instances are equally effected (dev, sandbox, /hb)
  • One of the theories is that there is race condition/resource blocking happening somewhere, perhaps on the shelves or in genome-related resources, though on one occasion GSuite parsing was timed and identified as taking very long time and we didn't find it access any shared resources
  • Another theory is that this is somehow related to Galaxy/how jobs are started
  • There is nothing exceptional found in logs except ProTo timeouts when slowness occurs
  • Sveinung: It is probable it could be something with the server and not software
  • Sveinung: It could be related to the forking of the process in ProTo
@sveinugu
Copy link
Member

Could also be (and perhaps most probable) something to do with the server setup, and not software related.

@sveinugu
Copy link
Member

@radmilak Any updates on the insilico monitoring software you queried about?

@radmilak
Copy link
Author

@sveinugu Some small updates - I found out I don't really have permissions to check the system logs on insilico so didn't find out anything by myself. But I contacted hpc-drift about this yesterday, haven't gotten any answer yet. Will post update here when they answer.

@radmilak
Copy link
Author

@sveinugu @morj-uio So, hpc-drift has restarted the monitoring client so it works again. They didn't say anything about why it was down but I guess if it would crash again we can ask them to investigate more then. It looks like this: Screenshot 2020-03-31 at 09.00.14.png

The monitoring pages are only accessible with an IP from the UiO range, so to do this from home you have to either tunnel via an UiO server (I also got this browser plugin so only grafana.uio.no requests use the tunnel) or set up a VPN but I didn't find much info about this on UiO website, so haven't tried. Or it should probably work with the UiO remote desktop thing.
To access it when you have the right IP, go to https://grafana.uio.no/ , click Home->host-resources and choose insilico.

@sveinugu
Copy link
Member

sveinugu commented Jul 7, 2020

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants