[Done v1.6.0] 1062:Duplicate entry for key 1 in db_cache

[lhungil]

I would prefer an INSERT ... ON DUPLICATE KEY ... for MySQL.

Longer syntax, but does performs an UPDATE if the insert fails versus always doing a DELETE + INSERT (if I remember correctly).

[RixStix]

After 10 days without generating the MyDebug file, I was beginning to think that all was fine though I did not schedule a PCI scan on the first of the month which historically I see the file generated many times during a scan.

Overnight, the 100K MyDebug file was generated.

[DrByte]

Just to close the loop here ...
We've opted for ON DUPLICATE KEY UPDATE for ZC v1.6.0:
https://github.com/zencart/zencart/pull/244/files
and
https://github.com/zencart/zencart/pull/245/files

(Text in red is stuff that was removed. Text in green is stuff that was added/changed.
Basically, remove all the lines that are shown as starting with a "-", and replace/add all lines that are shown as starting with a "+". The red and green are just clarifying guides.)

[RodG]

Just to close the loop here ...
We've opted for ON DUPLICATE KEY UPDATE for ZC v1.6.0:
https://github.com/zencart/zencart/pull/244
and
https://github.com/zencart/zencart/pull/245

I don't wish to be a pain, but is this really the best of the possible solutions?

What happens when two visitors get given the same sessionID, but only one can be entered into the database, wouldn't the "on duplicate key" effectively cause customer#2 to 'steal' the SessionID assigned to customer#1?

On the other hand, the 'insert ignore' will cause customer#2's insert to fail, which I assume will result in a new sessionID being generated for them, and customer#1's sessionID would remain unchanged.

I'm not seeing any benefits with the 'on duplicate key update' in this particular scenario.

Just my 2cents worth.

Cheers
RodG

[lhungil]

The place in sessions.php where the "ON DUPLICATE KEY UPDATE" occurs is only called at the END of a script or when PHP needs to write / update the existing session in the database. The same code block also updates the expiry information for the session). Validation and session creation happen earlier in the process.

The change in "sessions.php" basically replaces an existing "select; insert if not in database, update if in database" with a "insert in database, if cannot insert then update instead". By removing the first "select" and using the "on duplicate update", we remove a race condition - typically experienced when people use "multiple windows or tabs" in the browser.

[RixStix]

I may be misunderstanding something but there are 2 similar threads but offer different solutions.

1 edits sessions.php
1 edits cache.php

The edit to sessions.php did not appear to make any difference in the log files generated.
The edit to cache.php did have an effect. ie: the log files are no longer generated.

Can someone clear my clouded head? Are they intended to resolve different 1062: duplicate entry log files or are they 2 different attempts to resolve the same issue? Or are they both required to resolve the same issue?

[DrByte]

Similar issue, two different cases.

[swguy]

Just to close the loop here ...
We've opted for ON DUPLICATE KEY UPDATE for ZC v1.6.0:
https://github.com/zencart/zencart/pull/244
and
https://github.com/zencart/zencart/pull/245

Do you think this note warrants inclusion in the 1.5.4 fixes thread?
http://www.zen-cart.com/showthread.p...s)-with-v1-5-4

[DrByte]

Do you think this note warrants inclusion in the 1.5.4 fixes thread?
http://www.zen-cart.com/showthread.p...s)-with-v1-5-4

While it's not specific only to 1.5.4, I've posted it there.