Issue Details (XML | Word | Printable)

Key: CORE-3927
Type: Bug Bug
Status: Open Open
Priority: Major Major
Assignee: Unassigned
Reporter: Viktor Belzetskiy
Votes: 0
Watchers: 9
Operations

If you were logged in you would be able to see more operations.
Firebird Core

Restore's hung for big database.

Created: 19/Sep/12 09:48 AM   Updated: 20/Sep/12 04:59 PM
Component/s: GBAK
Affects Version/s: 2.5.0
Fix Version/s: None

File Attachments: 1. File db_create.7z (26 kB)
2. File db_hang.7z (353 kB)

Environment: Firebird CS 2.5.1 x86, Firebird SS 2.5.2.26538 x64, Windows2008 x86, Windows 2008R2 x64


 Description  « Hide
The DB volume is about 2 bln records, 1800 tables with 5400 indexes.

Restore's hung is happening while "gbak.exe -r -user SYSDBA -pass masterkey
-p..." executing if the key (-o "restore one table at a time") is not used.

We made a number of tries and found out that restore's hung is not depends
on:
- Firefird's architecture
- Database page size
- Windows architecture (x64, x86)
- Protocol (INET, XNET, service_mgr)

For same bug's playback you should act according to following sequence:
1. Decompressing db_hang.7z.
2. Execute script from file db_script.sql (execute time* ~16h, final
database size ~600 Gb, 7575522329 records)
3. Run Backup database "gbak -b..." (execute time* 6h, size fbk ~400 Gb)
4. Run Restore "gbak -r..." without key -o "restore one table at a time".

In our case the hung took place on 330 Gb file for "TEMP_TBL_1087"-table
(gbak: 176110000 records restored), after about 3 hours* of Restore's
executing.
* The executing time is specified for my hardware

 All   Comments   Change History   Subversion Commits      Sort Order: Ascending order - Click to sort in descending order
There are no subversion log entries for this issue yet.