In certain circumstances, vsmserver starts to clump session verifications together, which makes it launch many tasks at the same time. This initially leads to slowdowns for interactive clients (thinlinc-login, tlwebadm), but will finally lead to vsmserver crashing because we run out of low ports to run all these jobs at the same time.
(In reply to comment #0) > but will finally lead to vsmserver crashing because we run out of low ports > to run all these jobs at the same time. Bug 217.
(In reply to comment #0) > In certain circumstances, vsmserver starts to clump session verifications > together, which makes it launch many tasks at the same time. This initially > leads to slowdowns for interactive clients (thinlinc-login, tlwebadm), but > will finally lead to vsmserver crashing because we run out of low ports to run > all these jobs at the same time. A naïve solution to the session verifications clumping up could be to add a randomness to the standard 600 second interval. I still need to verify this theory though.
(In reply to comment #0) > will finally lead to vsmserver crashing because we run out of low ports to run > all these jobs at the same time. Also related: Bug 3876.
Bug 5488 is another related bug.
(In reply to comment #0) > In certain circumstances, vsmserver starts to clump session verifications > together, which makes it launch many tasks at the same time. This initially > leads to slowdowns for interactive clients (thinlinc-login, tlwebadm), but > will finally lead to vsmserver crashing because we run out of low ports to run > all these jobs at the same time. Bug 5451 describes one scenario that causes multiple jobs to be launched more or less simultaneously - a long-running task that's blocking vsmserver.
