Web control page stop button
Posted: Sun Nov 24, 2019 9:37 am
Good day all.
I'm (mostly) a debian-based (buster 10.2 now) folding work unit donor of years standing, thousands of units returned for substantial total score. Client 7.5.1 runs as a system service ofc, so there need be no interactive login for folding work units to launch and progress once system boots -- providing internet feed is live when required. My search for resolution or alternate means in docs and forums makes my google fu appear inadequate.
The difficulty is administrative: in order to have backups/snapshots as useful as possible, of the file system hosting the folding work queue, client must stop between WUs for snapshot, empty of ephemera, making restoring from snapshot later not disruptive. So, ability to finish the active unit without downloading a new WU assignment is necessary, not just wish list. It's not sensible having to manually time network connection so as to allow upload of finished WU results while defeating download attempts, which are very quick for a single unit assignment.
Unfortunately, the web control page (localhost:7396) Stop button, followed by "Finish up, then stop", is effectively ignored by FAH client. It's been so for years, perhaps only on debian based installations. By contrast, Stop Now behaves as expected. This bug remains unresolved or untracked even, as far as I can tell, and this would be not much trouble if only there were simple direct alternate means to bring about the same effect.
Access to protocol docs for telnet connection to client eludes me. Purposely I haven't pursued ugly hacks to make fah-control package install on buster, they're plainly not worth the grief. Though whatever that interface is doing to communicate to the client, informing it to disable download of next WU when progress reaches near-finished, that's something one should in principle be able to do directly from the terminal, with only ordinary system utilities? Not even needing a telnet session at port 36330? a certain signal sent to the client via kill?
Advice / perspective appreciated. Thanks for the timely attention-
I'm (mostly) a debian-based (buster 10.2 now) folding work unit donor of years standing, thousands of units returned for substantial total score. Client 7.5.1 runs as a system service ofc, so there need be no interactive login for folding work units to launch and progress once system boots -- providing internet feed is live when required. My search for resolution or alternate means in docs and forums makes my google fu appear inadequate.
The difficulty is administrative: in order to have backups/snapshots as useful as possible, of the file system hosting the folding work queue, client must stop between WUs for snapshot, empty of ephemera, making restoring from snapshot later not disruptive. So, ability to finish the active unit without downloading a new WU assignment is necessary, not just wish list. It's not sensible having to manually time network connection so as to allow upload of finished WU results while defeating download attempts, which are very quick for a single unit assignment.
Unfortunately, the web control page (localhost:7396) Stop button, followed by "Finish up, then stop", is effectively ignored by FAH client. It's been so for years, perhaps only on debian based installations. By contrast, Stop Now behaves as expected. This bug remains unresolved or untracked even, as far as I can tell, and this would be not much trouble if only there were simple direct alternate means to bring about the same effect.
Access to protocol docs for telnet connection to client eludes me. Purposely I haven't pursued ugly hacks to make fah-control package install on buster, they're plainly not worth the grief. Though whatever that interface is doing to communicate to the client, informing it to disable download of next WU when progress reaches near-finished, that's something one should in principle be able to do directly from the terminal, with only ordinary system utilities? Not even needing a telnet session at port 36330? a certain signal sent to the client via kill?
Advice / perspective appreciated. Thanks for the timely attention-