Update failed

[TheFerryman]

Error (mysql: MySQL server has gone away(2006) [DELETE FROM wp_dfr_xopt WHERE skey='update_lock'])
[Technical details...]
New site, store id 35287
Regards
Bruce

[TheFerryman]

Just received email
This is to inform you that the automatic update on your website (www.mysite.co.uk) has failed. Please see the error message(s) below:
ERROR:
-Store data cannot be read,
Technical details:
plugin/plugin:197 hostpage: disabled
plugin/plugin:139 view=none
app/tracker:61 track: no referrer
app/update:90 BEGIN: status=read
app/update:139 UPDATE FAILED
ERROR: err_read not_readable(122)

I should add that the site did download for the first time before the error and is working fine, showing products, etc.

[TheFerryman]

Should also say I moved the site to a different server, so it is possible the dns is not resolving fully just yet, would that make a difference?

Bruce

[stefan]

As mentioned here the most common reason for the MySQL server has gone away error is that the server timed out and closed the connection.

Not sure how large your store is but it's probably that MySQL times out during the update of your store.

If you continue to have this issue, you should probably ask your host why your MySQL server keeps going away.

You could also try Eric's suggestion in above thread.

[TheFerryman]

It is not a huge site, it seems to get stuck reading 6mb of 6.2 mb.
I'll check it again tomorrow.
Thanks for your help.
Bruce

[stefan]

Please try changing your Download packet size to 500 on this page: WordPress Admin Area > Tools > Datafeedr > Updates.

[TheFerryman]

Stefan

I tried an update earlier this morning and it went through fine. I tried another a short time ago and it failed. Then I tried your packet size suggestion and it worked again.

I am still thinking this may be a dns issue, which should resolve itself in a day or so. I'll report back whether the packet size fix is needed.

I am fairly sure this is not a Datafeedr problem.

Thanks again for your help and best wishes for 2012.

Bruce

[TheFerryman]

I am now having this problem with various sites. I have tried using Firefox and haven't recreated it in FF so far, so it could be a browser issue (IE9).

Regards

Bruce

[TheFerryman]

Ignore my last post, it was Firefox which failed to report the issue, I then got an email with the notification.

ERROR:
-Error (mysql: Table '(*mymysqltable*)_tmp' doesn't exist(1146) [INSERT IGNORE INTO wp_dfr_shop_products_tmp(id,net_id,net_code,networ k,merchant_id,merchant,name,slu]),
Technical details:
plugin/plugin:197 hostpage: disabled
plugin/plugin:139 view=product
app/tracker:61 track: no referrer
app/update:90 BEGIN: status=read
app/update:139 UPDATE FAILED
ERROR: mysql: Table '(*mymysqltable*)__tmp' doesn't exist(1146) [INSERT IGNORE INTO wp_dfr_shop_products_tmp(id,net_id,net_code,networ k,merchant_id,merchant,name,slu] (663)

The store id for that one is 35013

Regards

Bruce

[Eric]

Hi

These "Table '***_tmp' doesn't exist" errors happen when your /tmp directory gets pruned during an update. This is probably because your hosting will prune the /tmp directory after it reaches a certain size.

So, you need to set up another directory.

1. Create a new directory somewhere within your home directory. This could be within your public_html folder or above it.

2. Make sure that the directory is writable (ie. CHMOD 777)

3. Get the absolute path to this directory (not the relative path).

4. Enter the absolute path here: WordPress Admin Area > Tools > Datafeedr > Directories > Temporary Directory

5. Click the [Save] button and make sure you get the green check mark icon to appear. If it doesn't appear, that means either you entered the wrong path or the directory is not writable.

After you have all of this set up, your updates should run without a problem.

Eric

UPDATE 04/26/12: I've posted a video tutorial on how to easily perform these steps: http://www.datafeedr.com/videos?vide...rc=yt#videotop

[TheFerryman]

Eric

That sounds like a lot of work, changing every existing site and every future site.
Will you be doing a fix soon which will take care of this for us?
If so, will it just set up new sites correctly, or will it correct the error in existing sites?
Thanks
Bruce

[Eric]

Hi

There's no way for us to make those changes on your server for each of your sites.

The other option is to allocate more space in your /tmp directory so that updates don't fail and make sure it gets pruned less often.

Eric

[TheFerryman]

Can it not be done in the "set up all directories" function in Datafeedr, with "create directory if necessary" ticked?
I can't see any way to do it in cpanel and I don't have root access.
If we have to ask the server company to do it, they will probably need to know how big we want it?
If we add a new directory to each site, which the host would not be pruning, wouldn't it just get full eventually and give a different error?
How about a cron job to delete datafeedr update files daily?
Regards
Bruce
PS Just thought of an even better idea, add a line to the datafeedr update script to delete the temporary files from the tmp directory after the update completes, then the problem will never arise!

[Eric]

Hi

The files do automatically get deleted, but only after the update is complete.

Most hosts allocate very little storage to the /tmp directory... If you have one store which is larger than the size of your /tmp directory, this will continue to fail. So, you only need to set up new temporary directories for large stores.

Also, cache files are also stored in your /tmp directories so if you have turned ON caching, then that will also be taking up space.

Your host may be able to allocate more space to the /tmp directory or make a directory alias so /tmp actually points to another directory with much more storage. You could ask them about that.

Eric

[TheFerryman]

Good point about the caching taking up space. I like random selection on pages, so I am not benefitting much from caching anyway.
So, my first option will be to turn off caching for any site which gives the error. If that doesn't help, I'll try the other options.
Really useful discussion, thanks.
Bruce