Bug #45304 backup fails on slow connection
Submitted: 3 Jun 2009 13:31 Modified: 9 Jul 2009 13:14
Reporter: Boris Murashin Email Updates:
Status: No Feedback Impact on me:
Category:MySQL Administrator Severity:S2 (Serious)
Version:1.2.17 OS:Microsoft Windows (vista, win7)
Assigned to: CPU Architecture:Any

[3 Jun 2009 13:31] Boris Murashin
Backup fails on slow 512kbit (but stable) adsl connection for tables larger than ~1mb. Backup goes about 99% and then "Backup have been stopped" dialog appears (no matter one table or 100 tables in backup, always 99%, never in the middle). Backup file contains information for all tables, but for part of them (for larger tables) data is truncated. It restores normaly, but tables contains less rows than origin.

Same laptop on other connection takes backup with no problem.

I understand that it could be hard to repeat this bug, but there is no additional info about error from the program, neither log files found in program files. If you tell me how to get more details i could post them.

How to repeat:
try to take backup for large table on slow connection
[3 Jun 2009 15:15] Miguel Solorzano
Thank you for the bug report. Have you tested if there is packet loss in that connection i.e: with ping?. Thanks in advance.
[3 Jun 2009 16:36] Boris Murashin
can't ping: no direct access to HSP DB-server, ssh tunnel. In bash can't ping too, no suid.
Any other way to test connection?
does MySQL admin uses "SELECT * INTO OUTFILE..." or something? I could execute script to query server in the same way to see whats wrong.
[4 Jun 2009 12:43] Miguel Solorzano
Are you able to use the mysqldump command line client tool instead of MySQL Administrator?. So you can compare the results?. Thanks in advance.
[4 Jun 2009 14:04] Boris Murashin
results are the same, but now there is some info:

in separate shells i've executed:
ping "ssh-host" -l 1454 -t
mysqldump -uUSER -pPASS -routput.sql --log-error=ouput.err --ignore-table=DB_NAME.ONE_BIG_TABLE -v -P3305 DB_NAME (tunneled to localhost:3305)

ping started before mysqldump and stopped after, -l 1454 is used as reasonable MTU size for adsl. Pinged "ssh-host" in assumption that HSP inner network is fast and reliable (it is: any mysql query in bash fires rapidly and never fails)

ping results:

Packets: sent = 56, received = 56, lost = 0 (0% loss)
min = 99msec, max = 423 msec, mean = 162 msec

Connection is stable as i mensioned earlier...

mysqldump results:

-- Connecting to localhost...
-- Retrieving table structure for table SOME_TABLE...
-- Sending SELECT query...
-- Retrieving rows...
mysqldump: Error 1053: Server shutdown in progress when dumping table `SOME_TABLE` at row: 11809

Is it Server dropping slow client?
[9 Jun 2009 13:14] Miguel Solorzano
Looks like you have timeout problems..try to increase net_read_timeout and others timeouts if needed.
[9 Jul 2009 23:01] Bugs System
No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".