Emergency Backup BLAT Servers: Difference between revisions

From Genecats
Jump to navigationJump to search
(Adding quick command-line test example)
Line 74: Line 74:
   
   
mm10 blat should now be fixed on the original blat server,  be sure to go to the RR and do another blat test.
mm10 blat should now be fixed on the original blat server,  be sure to go to the RR and do another blat test.
===Test with hgServer===
A quick command-line test example:
<pre>
gfClient blatx.soe.ucsc.edu 17845 /gbdb/strPur4/  test.fa dnaTestOut.psl
</pre>


[[Category:Browser QA]]
[[Category:Browser QA]]

Revision as of 17:52, 25 January 2018

Overview

In case certain blat servers go down, there is a a backup blat server ("blatx") that has ports pre-configured to host the "top 5" blatted assemblies (hg38, hg19, hg18, mm10, mm9). If the current blat server(s) hosting any of these 5 assemblies goes down, manually switch to the specified fail-over server in the list below by updating the hgcentral.blatServers table. A common solution is for the admins to restart the blat server that has gone down.

List of current backup "blatx" server ports

  1. db: hg38
  2. host: blatx
  3. trans port: 17786 isTrans=1, canPcr=0
  4. untrans port: 17787 isTrans=0, canPcr=1


  1. db: hg19
  2. host: blatx
  3. trans port: 17778 isTrans=1, canPcr=0
  4. untrans port: 17779 isTrans=0, canPcr=1


  1. db: hg18
  2. host: blatx
  3. trans port: 17782 isTrans=1, canPcr=0
  4. untrans port: 17783 isTrans=0, canPcr=1


  1. db: mm10
  2. host: blatx
  3. trans port: 17780 isTrans=1, canPcr=0
  4. untrans port: 17781 isTrans=0, canPcr=1


  1. db: mm10
  2. host: blatx
  3. trans port: 17784 isTrans=1, canPcr=0
  4. untrans port: 17785 isTrans=0, canPcr=1

"mm10 blat is down!" Example

How is mm10 blat currently configured?

hgsql -h genome-centdb -e "select * from blatServers where db='mm10'" hgcentral
+------+--------+-------+---------+--------+
| db   | host   | port  | isTrans | canPcr |
+------+--------+-------+---------+--------+
| mm10 | blat1d | 17779 |       0 |      1 |
| mm10 | blat1d | 17778 |       1 |      0 |
+------+--------+-------+---------+--------+

Update the blatServers table to point mm10 to the backup "blatx" server:

hgsql -h genome-centdb -e "update blatServers set host='blatx' where db='mm10'" hgcentral
hgsql -h genome-centdb -e "update blatServers set port='17780' where db='mm10' and isTrans='1'" hgcentral
hgsql -h genome-centdb -e "update blatServers set port='17781' where db='mm10' and isTrans='0'" hgcentral

Check the change:

hgsql -h genome-centdb -e "select * from blatServers where db='mm10'" hgcentral
+------+-------+-------+---------+--------+
| db   | host  | port  | isTrans | canPcr |
+------+-------+-------+---------+--------+
| mm10 | blatx | 17781 |       0 |      1 |
| mm10 | blatx | 17780 |       1 |      0 |
+------+-------+-------+---------+--------+

Mm10 blat should now be fixed (is on the emergency server), go to the RR and do a blat test to be sure.

When the failed blat server has been fixed, roll back to it. In this example, move mm10 from blatx back to blat1d:

hgsql -h genome-centdb -e "update blatServers set host='blat1d' where db='mm10'" hgcentral
hgsql -h genome-centdb -e "update blatServers set port='17778' where db='mm10' and isTrans='1'" hgcentral
hgsql -h genome-centdb -e "update blatServers set port='17779' where db='mm10' and isTrans='0'" hgcentral

mm10 blat should now be fixed on the original blat server, be sure to go to the RR and do another blat test.

Test with hgServer

A quick command-line test example:

gfClient blatx.soe.ucsc.edu 17845 /gbdb/strPur4/  test.fa dnaTestOut.psl