After import of a specific mailbox, all search results and index entries disappeared

[Robert Christian Strasser]

Hello,

I've imported two mailboxes via IMAP pilerimport, after I imported the first one with only 111 emails, everything was fine and after the piler cron jobs I could see all messages in the Web GUI

pilerimport -i 127.0.0.1 -u mailbox1 -p password -x Trash,Spam,Drafts, ... processed: 111 [100%]

But when I imported the second mailbox via IMAP pilerimport, which has 406 emails, suddenly ALL search results disappeared from the the Web GUI

pilerimport -i 127.0.0.1 -u mailbox2 -p password -x Trash,Spam,Drafts, ... processed: 406 [100%]

If I run some simple MySQL commands to check, truly there are no more entries in the indexes in the database, but before the second import they actually where there, I have no idea why this is, I have even manually ran all the cronjob commands

Here are the MySQL commands I ran to check

select count(*) from sph_index;

returns 0 which before the second import returned 111

mysql -h 127.0.0.1 -P 9306

mysql> select * from main1,delta1;

returns Empty set (0.00 sec) after the second import

I have even completely deleted the piler database and sphinx indexes and started all over with a new piler database again, still only the first mailbox works.

I've also tried to update from 1.3.4 to the master branch version 1.3.5 pre1, and I still have the same issue.

The mail logs and the sphinx log which I included in sphinx.conf to get more infos show no issues, the index tasks seem to run through properly, but for no sane reason nothing is written in the database

I would be truly grateful for assistance,

Kindest regards, Robert Strasser

[Janos SUTO]

The sph_index table holds temporary entries only. The indexer.delta.sh script processes its content, and then clears them. Check if the metadata table has all the entries (111+406=517). I suspect that something has resetted (=emptied) the sphinx index files. Make sure that nothing touches them expect the piler cron jobs.

[Robert Christian Strasser]

Thank you for your prompt answer.

Actually new messages that arrived the last days which were routed via bcc copy appeared in the index. I still have no idea why the indexes were overwritten...

With your assistance I could reindex the "older" imported messages by getting their ids in the metadata table and then running the following command:

reindex -f 1 -t 517

The issue can be closed.

[Robert Christian Strasser]

all messages are now in the index