Dear All, 

I had a problem with automated backup. I run cron job to do backup at night everyday. But in 1 month, I often find 2-4 times my database couldn't success to backup at night. 

But at morning, sometimes I can backup it with no errors, sometimes I have to drop 1 table that makes problem first, so I can backup it. 

In firebird.log I see these messages :

1. At Friday 8 April 01:00, internal Firebird consistency check (decompression overran buffer (179), file: sqz.cpp line: 228)

In the morning 08:34, sweeper start and finished à after that I do backup, maybe sweeper fix it so I can do backup ?

2. Then on monday, 11 April at other database when I do query :

-  10:04 internal Firebird consistency check (cannot find record fragment (248), file: dpm.cpp line: 1246)

-  Looks like When I try to Gfix there is a lot of messages :

i. 11:48 Fragmented record 5312277 is corrupt in table JADWAL_DETIL2 (177)

ii. Index 2 misses node on page 20403 level 1 in table CPM (183) -- And several messages like that

iii. Page 27149 is an orphan – And 16 messages like that  (total = 17)

Info about my firebird server :

-  O/S : CentOS release 6.7

-  Firebird Classic Server 2.5.3 (force write enabled)

-  Application PERL. There is a long process to generate report that is often running. This process is deleting around 2000 records, and insert the new one again about 2000 records too. I had ever used "insert select" for fast performance, but now I used insert query one by one.  

Are there any clue why this often happen ? 

On the other case, firebird service is also often shutting down it self. I can see it from firebird.log :  Shutting down the server with 1 active connection(s) to 1 database(s), 0 active service(s)

Anybody had ever experience this too ?