[DAS2] DAS/2 Code Sprint, August 14-18

Chervitz, Steve Steve_Chervitz at affymetrix.com
Thu Aug 10 20:14:48 UTC 2006


The netaffxdas das/2 server is back up now. Turned out to be a memory
trouble. The server got some whomping queries thrown at it, such as these:

M_musculus_Aug_2005/features?overlaps=chr1/0:194923535;type=mrna;format=bps

H_sapiens_Mar_2006/features?overlaps=chr20/0:62435964;type=mrna;format=bps

Which it could not complete due to out of memory errors. But it could handle
this sizeable query even after the above failed:

H_sapiens_May_2004/features?overlaps=chr20/0:62435964;type=refseq;format=brs

Eventually, Jetty just decided it had enough and shut down it's connection,
shouting: WARN!! Stopping Acceptor ServerSocket

My fix was to restart the das/2 server giving the java process another 200M
of maximal heap. However, both das/1 and das/2 servers can now potentially
claim 89% of physical ram on that box, which could become unhealthy.

Ed notes that we might want to prevent such big queries in the first place.
There is an error code in the das spec for this (HTTP error 413 "Request
Entity Too Large"). But how do we determine the what's a reasonable maximum
allowable query result? It will depend on the feature density on a
particular assembly. This could be a good action item for the code sprint.

Steve



> From: "Helt,Gregg" <Gregg_Helt at affymetrix.com>
> Date: Thu, 10 Aug 2006 09:39:12 -0700
> To: Brian Gilman <gilmanb at pantherinformatics.com>
> Cc: DAS/2 <das2 at lists.open-bio.org>, "Chervitz, Steve"
> <Steve_Chervitz at affymetrix.com>
> Conversation: [DAS2] DAS/2 Code Sprint, August 14-18
> Subject: RE: [DAS2] DAS/2 Code Sprint, August 14-18
> 
> Apologies, it looks like we're currently having some problem with proxy
> redirection on the Affy DAS/2 server.  Steve, can you check on this?
> When I request anything but the top level ~/sequence, I'm getting back
> HTTP error 502 "Bad Gateway" with the message:
> "The proxy server received an invalid response from an upstream server."
> 
> However, I just tried the biopackages server and it is working, though
> response times are slower than usual (unless the response has already
> been cached).  Here's a feature query I recently ran, so it will be
> returned quickly from the server cache:
> 
> http://das.biopackages.net/das/genome/human/17/feature?overlaps=chr21/26
> 027736:26068042;type=SO:mRNA
> 
> hope that helps,
> gregg
> 
>> -----Original Message-----
>> From: Brian Gilman [mailto:gilmanb at pantherinformatics.com]
>> Sent: Thursday, August 10, 2006 8:10 AM
>> To: Helt,Gregg
>> Cc: DAS/2
>> Subject: Re: [DAS2] DAS/2 Code Sprint, August 14-18
>> 
>> Trying to get a features document?
>> 
>>     Hello Greg et al. I'm desperately trying to get a features
> document
>> out of one of the DAS 2 servers and have not been able to do it yet.
> Can
>> someone help me out!?
>> 
>>                                            Thanks!
>> 
>>                                               -B
>> 
>> Helt,Gregg wrote:
>> 
>>> Affymetrix is hosting a DAS/2 code sprint on August 14-18, to
> coincide
>>> with the CSB conference at Stanford.  The sprint will be held at
> Affy's
>>> Santa Clara location, which is about a 20 minute drive from the
> Stanford
>>> campus.  For those attending CSB, the proximity should make it easy
> to
>>> join in, even if it's just for a morning or afternoon.  We can
> provide
>>> transportation to and from CSB if needed.  If you are interested in
>>> attending please email me, and specify whether you'll need a
> workstation
>>> or will be bringing your own laptop.
>>> 
>>> This is a code sprint, so the focus will be on DAS/2 client and
> server
>>> implementations.  As with previous sprints I'd like to start each day
>>> with a teleconference at 9 AM Pacific time.  If you can't be there
>>> physically but still want to participate, please join in!
>>> 
>>>            Gregg
>>> 
>>> 
>>> _______________________________________________
>>> DAS2 mailing list
>>> DAS2 at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/das2
>>> 
>>> 
>>> 
>>> 
> 





More information about the DAS2 mailing list