Improve speed when recovering from a poller from offline state

[bmfmancini]

Hey Guys

I was testing a failure scenario in the Main poller was taken down for a while
came back restored the main poller the remote poller of course was in heartbeat status

it showed that is had 5M offline records the log showed poller_output_table full
but it appears the recovery script was not being initiated by cacti in the background even though the main database was restored

I manually ran the poller_recovery.php script and it has been running for almost 1 hour seems to be running pretty slow my lab only has 130 devices on it so I could imaging if this was in a production system with tons of devices that it would take a huge amount of time

I am not sure if this is typically slow as I have not had to run this manually before

Thanks!

[TheWitness]

I just looked at the code. I have to talk to Jimmy about his drinking habit. Just kidding. I've fixed it up. Have an upload in a bit.

[TheWitness]

Not sure you want to test this, but it'll be way faster now.

[TheWitness]

My systems was able to do ~2000 in about 1.7 seconds. So, 5M should take no more than a minute.

[bmfmancini]

wow that is an improvement I tested the old script vs the new similar failure scenario and notice a big performance boost

Thanks !

[TheWitness]

Okay, close if happy.