Re: [TOOLS-DEVELOPMENT] Narrowing the slowdown down...

[Robert Sparks <[e-mail]>]

Hi Magnus -

Thanks for helping brainstorm. In this case, I'm pretty confident that the history of the document has no impact on the bug.

RjS

On Jun 27, 2011, at 10:19 AM, Magnus Westerlund wrote:

> Hi Glen,
> 
> This might help you.
> 
> This document is from AVT that was concluded and then split into 4 WGs
> and the documents got re-assigned to the new WGs. I have earlier filed
> several ticket on the WG tracker regarding documents that has not
> correctly accessible despite it showing up as WG document. Apparently
> there is more than one field that contains which WG the document belongs
> to. Henrik should know more about this issue.
> 
> So it is not surprising if there are issues around AVTs old documents as
> they have been re-assigned.
> 
> Cheers
> 
> Magnus
> 
> 
> On 2011-06-27 16:44, Glen wrote:
>> All -
>> 
>> I have sent detailed data to the tools team and Henrik, but I wanted to alert
>> everyone to a pattern I've seen during my analysis:
>> 
>> This request:
>> 
>> POST /doc/draft-ietf-payload-rfc3016bis/edit/position/ HTTP/1.1" 302 - 
>> https://datatracker.ietf.org/doc/draft-ietf-payload-rfc3016bis/edit/position/
>> 
>> was seen in the logs at the start of both slowdowns, and I now suspect that
>> there may be database corruption and/or some problem with the code related
>> either to ballot positions generally, or this draft specifically.
>> 
>> It comes to my mind that, while I was gone, a request came in to clear the
>> ballot positions for a draft, which the secretariat did.  This may have been
>> the draft that was cleared - and clearing it may have caused some type of
>> problem for the datatracker.
>> 
>> Of course, the datatracker should not loop or fail even if data is bad, but
>> not all possibilities can be forseen.
>> 
>> It is my hope that we will both be able to correct a potential database
>> problem, and find and harden a potential datatracker bug, quickly.
>> 
>> In the meantime, until we hear from the tools team, it might be best to
>> at least refrain from voting on the above draft, if not all drafts.
>> 
>> If you do vote on a draft, and get a response, don't get too excited either
>> way.  The server actually survives for an hour or more once the bug starts
>> using resources (I'm actually proud of this - it's a HUGE server with lots
>> of resources - the old servers would have died much more quickly. ;-) so
>> things can appear okay for a while.
>> 
>> Now that we know what to look for, we can catch it earlier, but I'm still
>> hopeful for a quick fix and repair today.
>> 
>> Thanks,
>> Glen
>> 
> 
> 
> -- 
> 
> Magnus Westerlund
> 
> ----------------------------------------------------------------------
> Multimedia Technologies, Ericsson Research EAB/TVM
> ----------------------------------------------------------------------
> Ericsson AB                | Phone  [phone]
> Färögatan 6                | Mobile [phone]
> SE-164 80 Stockholm, Sweden| mailto: [e-mail]
> ----------------------------------------------------------------------
>