狮身人面像重新启动搜索恶魔问题以重新索引Delta索引

Question

I’m working on my application search engine. This search engine is using sphinx for fulltext searching.

I created a cron job which is responsible for updating indexes. Currently it is set to execute once in a day.

Here what im doing in this cron.

   indexer --all --rotate > sphinx.log

Sometimes the whole search module goes down and it shows this error on client application.

Notice (1024): Search query failed: connection to 127.0.0.1:9312 failed (errno=111, msg=Connection refused) [APP/models/behaviors/sphinx.php, line 134]

And on sphinx.log I found this error

FATAL: failed to lock /usr/local/sphinxsearch/delta_users/delta_users.spl: Resource temporarily unavailable, will not index. Try --rotate option.

It seems like 9312 is refused to connect, probably this is because the resource files are used by some other process. This is random behavior yet.

The environment is deployed on Linux (ubuntu) and application is created on cakephp. I’m trying to identify what cause it stops from restarting sphinx service again. I don’t know either this is sphinx configuration issue or some I/O or mysql problem. I think delta indexes causing some problem only.

Thanks in advance.

Answer 1

The problem is actually when it rotates all indexes i.e main and delta indexes altogether it locks the resource file of delta indexes. I moved the rotation of delta indexes with other cron job which is responsible for merging delta indexing.

indexer   delta_users delta_users_type
indexer --merge users  delta_users 
indexer --merge users_type delta_users_type

And rotate only main indexes separately on different cron

indexer --rotate -- users  users_type

instead of

indexer --rotate -- all 

Hence the separate rotation of Delta and Main index is the resolution.

Thanks