"mysql Server Has Gone Away" Still In 6.12

[Marjolein Katsma]

I had hoped the old “MySQL server has gone away” problem would have been tackled with the 6.12 version but if it was tackled it wasn't solved.

It just seems I cannot copy more than a few thousand rows in a table from one server to another. SQLyog starts up with 4 or 5 copies for plink.exe, but pretty soon they either go away one by one until none is left (at which point we get the “MySQL server has gone away” error, or – and this is a new one to me! – at the point when one of the plink.exe copies goes away (leaving one or more still active) the program reports 'Copy successful' and stops copying, when in reality only a part (or a fraction) of the rows have been copied.

It seems I just cannot get some tables copied – just when I urgently need to build a mirror of one of our workgroup's websites. The MySQL version on the target server is not the same as on the source and not under my control (we're just borrowing emergency space here).

Any suggestions or solutions?

[peterlaursen]

You are right – there is no change in this respect.

I still do not know if we can 'reinstantiate' PLINK if connection is lost between PLINK and he SSH Daemon.

mybe on Monday somebody else can tell if there has been considerations in this. The issue is recorded in our issue-tracker.

[Marjolein Katsma]

Thanks, Peter.

peterlaursen wrote on Nov 17 2007, 12:42 PM:

You are right – there is no change in this respect.

I still do not know if we can 'reinstantiate' PLINK if connection is lost between PLINK and he SSH Daemon.

I think this should be possible – in fact, I've been watching Process Explorer closely while SQLyog is doing something and I sometimes see an “extra” plink.exe appearing briefly near the end of a task, so 1) the program can obviously detect a missing connection (“MySQL server has gone away”) and 2) can clearly also start a new instance of plink.exe. To me that suggests a solution should be possible.

peterlaursen wrote:

mybe on Monday somebody else can tell if there has been considerations in this. The issue is recorded in our issue-tracker.

The 'Copy successful' problem (a different face of the same underlying problem?) is new though – and if anything more serious: if you don't go and check number of rows, you'd not even have a clue that the copy was not complete. I realised only later, after running into this with a few larger tables where the copy finished way too soon, that one of the first tables I copied – a small one – was also incomplete.

Please make sure this info is also added to the issue!

Thanks,

Marjolein

[Author]

Posts