How gracefully restart Sphinx search daemon after reindexing

Question

I've reindexed my Sphinx search with /usr/local/sphinx/bin/indexer --all --rotate and renamed my original index output files to something else. Simply changing the index argument passed to $sphinx->Query($query, $index); returns no results.

I suspected the daemon doesn't know the new index files exist. So I ran

sudo /usr/local/sphinx/bin/searchd

again to try to restart it. But it threw

FATAL: failed to lock pid file '/usr/local/sphinx/var/log/searchd.pid': Resource temporarily unavailable (searchd already running?)

I had to kill the 2 processes of the search daemon and start it again to grab from the new index files. Is there a graceful way to restart it?

score 30 · Accepted Answer · answered Nov 30 '11 at 12:34

30

I know this is a late answer, but just so you know, to 'restart' Sphinx, you need to stop it then start it (as in, two distinct processes).

To stop it, call searchd --stop then just start it again with searchd.

answered Nov 30 '11 at 12:34

Narcissus

3,144
3
27
40

2

Imho `/etc/init.d/sphinxsearch restart` is better – pevik Apr 11 '13 at 11:24
5

The command `/etc/init.d/sphinxsearch restart` appears to be distribution-specific. – Diogenes Creosote Nov 25 '14 at 23:41
Building on @Narcissus answer, you can wrap up both calls in a shell script – Hatem Jaber Jan 19 '16 at 19:29

daxroc · Answer 2 · 2011-10-27T22:43:57.817

You'll need to call indexer on the new index to create it and then --rotate to update it.

So it would be something like

indexer --config /path/to/config.conf indexname

And then when you just want to update your indexes

indexer --config /path/to/config.conf --rotate --all

This will create a temporary copy of each index and replace the old ones when finished. For more info on what actually happens see http://sphinxsearch.com/docs/manual-0.9.9.html#ref-indexer

On the other error your getting Do

ps aux | grep searchd

if it returns no results, then remove /usr/local/sphinx/var/log/searchd.pid and start searchd again

score 4 · Answer 3 · edited May 23 '17 at 23:36

4

It seems there is an issue with the searchd --stop command failing to stop the daemon on some instances of Sphinx.

Try: service sphinxsearch stop

See: https://bugs.launchpad.net/ubuntu/+source/sphinxsearch/+bug/990395

edited May 23 '17 at 23:36

d-_-b

21,536
40
150
256

answered Nov 14 '12 at 22:44

y_s

138
4

In my case I run into this because of having nonstandard pid file in `/etc/sphinxsearch/sphinx.conf` - which was different from `/etc/init.d/sphinxsearch`. – pevik Apr 11 '13 at 11:33

score 0 · Answer 4 · answered May 23 '17 at 23:42

0

service searchd start worked for me on CentOS

answered May 23 '17 at 23:42

d-_-b

21,536
40
150
256

How gracefully restart Sphinx search daemon after reindexing

4 Answers4