Updating blat servers: Difference between revisions
From genomewiki
Jump to navigationJump to search
Marygoldman (talk | contribs) (updating with Brooke's suggestions. Mainly that we stop testing beta so throughly) |
Marygoldman (talk | contribs) (adding information about new blat servers and where they are located) |
||
Line 1: | Line 1: | ||
There are 3 places where blat servers are located. Assemblies that are important (eg. human or mouse) and are expected to get a lot of hits are hosted here at UCSC and will have a blat host name like this: blat# (where # is a number). Assemblies that are less important are hosted in the cloud and will have a blat host name like this: cloudblat1.cse.ucsc.edu. Assemblies that are only on hgwdev will get a blat server also at UCSC, but in a less important room. They will have a blat host name like blatX (where X is a letter). Wherever the blat server is hosted, the steps below will allow you to transfer the pointers to a new machine. | |||
* Do a select statement on dev to make sure that everything is where you expect it to be: | * Do a select statement on dev to make sure that everything is where you expect it to be: | ||
<pre> | <pre> |
Revision as of 18:49, 22 October 2010
There are 3 places where blat servers are located. Assemblies that are important (eg. human or mouse) and are expected to get a lot of hits are hosted here at UCSC and will have a blat host name like this: blat# (where # is a number). Assemblies that are less important are hosted in the cloud and will have a blat host name like this: cloudblat1.cse.ucsc.edu. Assemblies that are only on hgwdev will get a blat server also at UCSC, but in a less important room. They will have a blat host name like blatX (where X is a letter). Wherever the blat server is hosted, the steps below will allow you to transfer the pointers to a new machine.
- Do a select statement on dev to make sure that everything is where you expect it to be:
hgwdev: hgsql hgcentraltest mysql> select * from blatServers where host="blat#" ORDER BY port; +---------+--------+-------+---------+--------+ | db | host | port | isTrans | canPcr | +---------+--------+-------+---------+--------+ | apiMel1 | blat12 | 17778 | 1 | 0 | | apiMel1 | blat12 | 17779 | 0 | 1 | | monDom1 | blat12 | 17786 | 1 | 0 | | monDom1 | blat12 | 17787 | 0 | 0 | | caeRei0 | blat12 | 17790 | 1 | 0 | | caeRei0 | blat12 | 17791 | 0 | 1 | +---------+--------+-------+---------+--------+
- Get DNA and protein sequences for a gene for given assembly and paste below for future reference:
- DNA
- protein
- Open 6 windows: BLAT DNA, BLAT protein, PCR for dev and the RR (we don't check on beta since it doesn't really matter if blat is working there or not). Make sure that all 6 windows function without errors.
Getting into hgcentral database on dev, beta, RR: DEV: hgwdev: hgsql hgcentraltest BETA: hgwbeta: hgsql hgcentralbeta RR: hgwdev: hgsql -h genome-centdb hgcentral Statements for changing pointers: update blatServers set host='blat#', port=177## where db='xxxXxx#' and isTrans=1; update blatServers set host='blat#', port=177## where db='xxxXxx#' and isTrans=0; **Note that the admins will simiply say: bosTau3 trans port 17786 untrans port 17787 This means put the trans port number for isTrans=1 and the untrans port number for isTrans=0
- Go to hgcentral on dev and change the pointers for the assembly of interest.
- Go back to the browser, hit back for the 3 windows open on dev, and then click re-submit. Check to see if broken. If not, go on; if yes, troubleshoot until solution is found.
- Repeat step 4 for beta and 4-5 for the RR.
- Rebuild hgCentral.sql for mirror sites and push to hgdownload:
- cd $WEEKLYBLD
- Run buildHgCentralSql.csh real (can run w/o 'real' to just see diffs). hgcentral.sql will be automatically pushed to hgdownload.