cycledash
cycledash copied to clipboard
hammerlab
Annotations column appears and disappears non-deterministically
Appears to be there on (at least some) old runs (e.g. http://cycledash.demeter.hpc.mssm.edu/runs/270/examine?query=ORDER+BY+contig%2C+position), but isn't running anymore? Is the worker queue up still @danvk? It must be if extractions are happening; could be the annotator is downloading a new dataset, and it'll take a while after redeploy for it to complete its first annotation…
The gene_names were there before I refreshed to the latest deploy, this was submitted before the update.
So I did have them in the new deploy in another tab, refreshed that tab and now they are gone.
...and now gone: http://cycledash.demeter.hpc.mssm.edu/runs/327/examine?query=sample_name+%3D+TUMOR+ORDER+BY+contig%2C+position
Did you guys figure out what was going on last night? I'm seeing an annotations column on every examine page linked to in this issue.
I do see cycledash-worker tasks running:
$ sudo initctl list
cycledash-worker (3) start/running, process 18407
cycledash-worker (2) start/running, process 18402
cycledash-worker (1) start/running, process 18398
Try refreshing the page a bunch; its appearance is totally random.
On Tue, Jun 16, 2015 at 6:48 PM, Dan Vanderkam [email protected] wrote:
Did you guys figure out what was going on last night? I'm seeing an annotations column on every examine page linked to in this issue. I do see cycledash-worker tasks running:
$ sudo initctl list cycledash-worker (3) start/running, process 18407 cycledash-worker (2) start/running, process 18402 cycledash-worker (1) start/running, process 18398
Reply to this email directly or view it on GitHub: https://github.com/hammerlab/cycledash/issues/740#issuecomment-112494659
Yeah, quite :) I'm more than happy to look at this when I have a few hours. Haven't seen it locally, so it may need to wait until I'm back unless one of y'all get to it.
On Tue, Jun 16, 2015 at 6:51 PM, Dan Vanderkam [email protected] wrote:
I shift-reloaded Arun's page a few times and saw it. Fun!
Reply to this email directly or view it on GitHub: https://github.com/hammerlab/cycledash/issues/740#issuecomment-112495308
Yeah my guess is this could be related to the race conditions in RecordStore.
On Tue, Jun 16, 2015 at 6:53 PM, Dan Vanderkam [email protected] wrote:
I'm not seeing anything in the flask logs.
Reply to this email directly or view it on GitHub: https://github.com/hammerlab/cycledash/issues/740#issuecomment-112495784
I'm not sure how, though. I'd bet the backend is functioning just fine though.
On Tue, Jun 16, 2015 at 6:53 PM, Dan Vanderkam [email protected] wrote:
I'm not seeing anything in the flask logs.
Reply to this email directly or view it on GitHub: https://github.com/hammerlab/cycledash/issues/740#issuecomment-112495784
Looking at the network traces supports that theory: even when the annotations column isn't visible, I see it mentioned both in the examine HTML and in the /genotypes response.
The store needs a total rewrite to be safe anyway. This may be the forcing function.
On Tue, Jun 16, 2015 at 6:59 PM, Dan Vanderkam [email protected] wrote:
Looking at the network traces supports that theory: even when the annotations column isn't visible, I see it mentioned both in the examine HTML and in the
/genotypesresponse.Reply to this email directly or view it on GitHub: https://github.com/hammerlab/cycledash/issues/740#issuecomment-112497125