Opened 14 years ago
Closed 14 years ago
#8790 closed defect (Won't Fix)
Patch to fix eta timer bug in mencoder jobs
Reported by: | Owned by: | beirdo | |
---|---|---|---|
Priority: | minor | Milestone: | 0.24 |
Component: | Apps - Nuvexport | Version: | Master Head |
Severity: | low | Keywords: | nuvexport, patch, eta |
Cc: | Ticket locked: | yes |
Description
In Ticket #8788 from yesterday I posted an enhancement that now allows mencoder jobs to show an eta while in progress.
While doing some live environment work last night I found a bug in the code related to the timer handling (specifically the death timer), I believe this is related to the total frames count values being mismatched in what mythtranscode expects and what the file seems to have, I will do some further investigation to get to the bottom of it.
The bug manifests as either a mythtranscode died early error on a single pass or a crash in a two pass nuvexport --mencoder job.
To apply a temporary fix I have adjusted the death watch timer in this attached patch until I can fully understand what's causing the bug.
Attachments (1)
Change History (12)
Changed 14 years ago by
Attachment: | mencoder.pm.diff added |
---|
comment:1 Changed 14 years ago by
comment:2 Changed 14 years ago by
Owner: | changed from xris to beirdo |
---|---|
Status: | new → assigned |
comment:3 Changed 14 years ago by
Resolution: | → worksforme |
---|---|
Status: | assigned → closed |
This works just fine for me. There is likely something funky in your setup, and increasing the death timer is certainly not something I'm looking at implementing in nuvexport at large. It is likely needed for you to debug on your end, but will not be committed.
I would suggest the use of nuvexport --debug, and then running the commands separately to see what barfs.
comment:4 Changed 14 years ago by
Resolution: | worksforme |
---|---|
Status: | closed → new |
Hi
OK I've attached two screengrabs from last night of the same file being nuvexported.
One is the failed pass and the the successful pass after I changed the deathtimer.
If you look at the pass that worked you can see the framecount is higher then the total number of frames at the end, that doesn't make sense.
My setup is the machines running the nuvexport job pulls the file from the backend via an NFS share, when I tried to do a two pass job it crashed exactly between the first and second pass (indicative of the same problem above), changing the deathtimer from 30 seconds to 180 allowed both a single pass and a two pass job to work OK. As the deathtimer is only a 30 second counter is it a major issue to change it to 180 seconds? I'll try to get to the bottom of the frame counter error and see if I can give you more feedback.
I'm happy for you to leave this as a closed ticket, however if people raise an issue please be aware of this.
comment:5 Changed 14 years ago by
Two screengrabs are here.
comment:6 Changed 14 years ago by
Status: | new → assigned |
---|
comment:7 Changed 14 years ago by
Priority: | blocker → minor |
---|---|
Severity: | medium → low |
Status: | assigned → infoneeded |
You need to do some digging. As I said before, I'd like to see you run nuvexport with --debug, and manually run the steps. This may show out the underlying error.
This is *not* a blocker bug. Nuvexport works for other people just fine, and continues to work fine. I may at some point need a sample file that it fails on for you to test on my setup too.
I'll look over what you added as well, but I don't expect to find anything. 30 seconds should be an adequate timeout, but I'll see when I read through the section of code again.
comment:8 Changed 14 years ago by
Milestone: | 0.24 → unknown |
---|---|
Status: | infoneeded → assigned |
comment:9 Changed 14 years ago by
Status: | assigned → infoneeded |
---|
comment:11 Changed 14 years ago by
Milestone: | unknown → 0.24 |
---|---|
Resolution: | → Won't Fix |
Status: | infoneeded → closed |
Ticket locked: | set |
the email address in the original posting is wrong, sorry typo.