mail missing in webui

Issue #631 closed
Crzaccount created an issue

Hi Janos,

      In metadata, the total mail count is approx 1.3 crore. But when we login as auditor the total mail count is 1.06crore. It seems some of the mails are not displaying in webui.

We are having approximately 30 used indices. Pls let us know how to solve this issue apart from reindex -a.

Also please let us know the below possibilities

  • reindex only the specific recipient mail.
  • reindex the mails between specific range date.

Comments (23)

  1. Crzaccount reporter

    piler -v o/p as follows

    ###Start [root@archival ~]# piler -v piler 1.1.1, build 904, Janos SUTO sj@acts.hu

    Build Date: Tue Oct 6 12:54:08 NPT 2015 ldd version: ldd (GNU libc) 2.17 gcc version: gcc version 4.8.3 20140911 (Red Hat 4.8.3-9) (GCC) Configure command: ./configure --localstatedir=/var --with-database=mysql [root@archival ~]# ###End

    If you need any other information, pls let us know.

  2. Janos SUTO repo owner

    I'm not sure what a crore is, but it's pretty unusual to use 30 indices. How big are your indices, and how many emails do you have in the archive? How much memory do you have? Please use international units.

  3. Crzaccount reporter

    In metadata the total mail count is 13459348. But when i logged in as auditor, the total mail count is 10669120

    I'm having 32GB RAM. piler is installed on CentOS 7.1 with 64bit. My indices maximum size is 3.7GB.

    Do you have any details?

  4. Janos SUTO repo owner

    OK, you have to check all your indices if combined they provide a continuous index. That is run the following queries:

    select min(id), max(id) from main1;
    select * from main1;
    show meta like 'total_found';
    

    query 1 shows the starting and ending serial id for the given index query 2 is just to pull some values query 3 show the total number of indexed documents for the given index

    Now perform these queries for all indices, and verify that they contain all your documents. Also check if you have ever purged any email from the archive:

    select id from metadata where deleted=1;
    
  5. Janos SUTO repo owner

    Check out, if the sphinx files have reasonable sizes. Also make sure that no indexer related stuff runs from a cron job other than piler's crontab.

  6. Janos SUTO repo owner

    What's happened since it was working properly? Something has just resetted the sphinx database for sure.

  7. Matt

    I know from the beginning I had issues with searching, on 1.1.1, never got into that much as piler itself was collecting the emails well (following the DB).

    I don't have much data in this setup so reinitialize would be possible if possible.

    What can cause a reset of the sphinx database ?

  8. Janos SUTO repo owner

    Eg. running "indexer --all". I suggest you to reindex all emails, check out the docs on how to do that.

  9. Matt

    I did the reindex and it indexed indeed messages but no older emails in my webgui.

    It also seems that the searchdomains, etc are not honored per users. I mean here that all is shown not the domains you put in there.

  10. Janos SUTO repo owner

    It takes <30 mins until emails appear in the gui. Also check if the sphinx files have some reasonable sizes. Also show me the search query you can see in the mail log.

    The searchdomains issue needs some elaboration: how it's not honored? For which user? What is shown what shouldn't be? Open a new issue for this problem, and elaborate it.

  11. Crzaccount reporter

    Hi Janos,

          all indexes are fine. i.e. it is combined with that previous index. Pls find the o/p
    

    ###Start MySQL [(none)]> show meta like 'total_found'; +---------------+--------+ | Variable_name | Value | +---------------+--------+ | total_found | 981275 | +---------------+--------+ 1 row in set (0.02 sec)

    MySQL [(none)]> ###End

    Pls let me know what is to be done further. Also kindly let me know is there any possibility for the below options

    1- reindex only the specific recipient mail. 2- reindex the mails between specific range date.

  12. Crzaccount reporter

    I have found the id from metadata for specific date based on the arrived column.

    ##Start select id from metadata where arrived>=1431019813 and arrived<=1432841413 order by id asc limit 10; select id from metadata where arrived>=1431019813 and arrived<=1432841413 order by id desc limit 10; ###End

    I'm doing the reindex for these ids.

  13. Crzaccount reporter

    Hi Janos,

            When i login as auditor & search the mails, i found that most of the mails are "Message Failed Verification".
    

    I got the below log when I search in webui.

    ###Start Nov 12 16:10:00 archival pilerget[7936]: /var/piler/store/00/550/ad/97/400000005500b510088ef8ac003ddafbad97.m: cannot open() Nov 12 16:10:00 archival pilerget[7938]: /var/piler/store/00/550/ad/97/400000005500b510088ef8ac003ddafbad97.m: cannot open() Nov 12 16:10:00 archival pilerget[7940]: /var/piler/store/00/550/8c/2c/400000005500b4681e1dc52c00e5a3ac8c2c.m: cannot open() Nov 12 16:10:00 archival pilerget[7942]: /var/piler/store/00/550/8c/2c/400000005500b4681e1dc52c00e5a3ac8c2c.m: cannot open() ###End

    The same kind of log is coming for attachments also.

    ###Start Nov 12 16:18:10 archival pilerget[8636]: /var/piler/store/00/54a/a4/f6/4000000054aca267262a3da400b3c405a4f6.a1: cannot open() Nov 12 16:18:10 archival pilerget[8636]: /var/piler/store/00/54a/a4/f6/4000000054aca267262a3da400b3c405a4f6.a2: cannot open() ###End

    I have noticed that store directory doesn't have the folder called "530" & "54a". I couldn't understand why it is looking the non existing directory. Pls find the directories present in store

    ###Start [root@archival ~]# ls /var/piler/store/00/ 52c 52e 531 535 537 539 53c 53e 541 545 547 54b 54d 54f 552 554 556 558 55a 55c 55e 561 563 52d 52f 533 536 538 53b 53d 540 542 546 549 54c 54e 551 553 555 557 559 55b 55d 55f 562 564 [root@archival ~]# ###End

    Pls let me know how to rectify the issue. Since most of the mails are like this, we couldn't restore this mail. We need to fix this issue as soon as possible.

    Pls help me.

  14. Janos SUTO repo owner

    Show me the following output:

    select id, sent, arrived, archived, deleted from metadata where piler_id='400000005500b510088ef8ac003ddafbad97';
    
  15. Crzaccount reporter

    Pls find the o/p. archived column not exists in metadata. So i have excluded that from the sql query.

    ###Start MariaDB [piler]> select id, sent, arrived, archived, deleted from metadata where piler_id='400000005500b510088ef8ac003ddafbad97'; ERROR 1054 (42S22): Unknown column 'archived' in 'field list' MariaDB [piler]> select id, sent, arrived, deleted from metadata where piler_id='400000005500b510088ef8ac003ddafbad97'; +---------+------------+------------+---------+ | id | sent | arrived | deleted | +---------+------------+------------+---------+ | 8462803 | 1426114798 | 1426109702 | 0 | +---------+------------+------------+---------+ 1 row in set (0.00 sec)

    MariaDB [piler]> ###End

  16. Janos SUTO repo owner

    You are right, I meant 'arrived'. However, based on the sent value, the email was sent on 1970.04.08. Not sure if it's right. Anyway try looking inside /var/piler/store/00/fb/ad/97 directory. Let me know if you can find it there or not.

  17. Crzaccount reporter

    I couldn't find out the directory. /var/piler/store/00/fb/ad/97

    In /var/piler/store/00 the below directories only there.

    ###Start 52c 52e 531 535 537 539 53c 53e 541 545 547 54b 54d 54f 552 554 556 558 55a 55c 55e 561 563 52d 52f 533 536 538 53b 53d 540 542 546 549 54c 54e 551 553 555 557 559 55b 55d 55f 562 564 ###End

    Pls let me know, from which path i need to find.

  18. Janos SUTO repo owner

    If you don't have such a directory, and not even directly under /var/piler/store, then I have no idea where the mail is.

  19. Log in to comment