Hmm - bit weird. If you zipped up the old $ZENHOME/perf/Devices and restored on the new build then you ought to end up with the same set of stuff that you should have if you had done the backup including the perfdata. Can you pinpoint what is missing?
From memory, if you do a zenbackup including the perfdata (and the zodb as before), then a restore over the top of what you have now should be OK (though I would definitely have a backed up / snapshotted recovery position.
Subject: |
RE: Zenoss 4.2.5 auto deploy script feedback |
Author: |
Clayton Tavernier |
Posted: |
2018-01-09 19:03 |
I didn't look too close but I'm thinking I didn't zip the empty sub-folders.
I'm taking a new backup. "zenbackup --no-eventsdb --no-save-mysql-access".
And it failed with:
2018-01-09 14:05:56,008 CRITICAL zenbackup: 'Performance Data backup failed.'
2018-01-09 14:05:56,008 CRITICAL zenbackup: 'Backup packaging failed.'
And now I remember that I left out the perf data before because of the 32-bit to 64-bit issue but I'm wondering what's the point of backing up without the perf data? I mean, it looks like I got the list of devices moved from the old server to the new server but without the data? Is there a way to get Zenoss to use the device list to go out and "discover" them?
Taking a new backup. "zenbackup --no-eventsdb --no-perfdata --no-save-mysql-access".
I'm also zipping up the Devices folders from the old server. "tar czvf devices.tgz /opt/zenoss/perf/*". I wasn't sure if I should grab just Devices or should I include the other files/folders from perf. Now I'm thinking just Devices.
I took a closer look at the two Devices folders and it just looks like there are folders missing from the new server. Weirdly, I can't find the original tar command on the old server to see what I did wrong.
So I did the tar -czvf on on the old server and the tar -xzvf on the new server and the totals are closer.
old_server# ls -la /opt/zenoss/perf/Devices
total 1692
new_server# ls -la /opt/zenoss/perf/Devices
total 1684
And I'm still getting the Zenoss error on the new server when I try to access device details.
------------------------------
Clayton
Network Computer Technician
------------------------------
Subject: |
RE: Zenoss 4.2.5 auto deploy script feedback |
Author: |
Jane Curry |
Posted: |
2018-01-15 06:26 |
Not quite sure what is failing now? Is it looking at performance data for a device?
You are correct that there is an issue if you are moving from a 32-bit to a 64-bit implementation. Forgotten the details but fundamentally, as you say, you can't see the data - subtly different format so either tar'ing up the old files or using zenbackup with perfdata, will not work. I vaguely remember a discussion about there being rrd export / import tools that might get around this but I'm afraid I never tried them.
Many people who go through the sort of rebuild process you are doing, decide that they will move the devices but abandon the old events and perf data. I'm sorry, but if you have gone from 32-bit to 64-bit then I have a nasty feeling that you won't get your perf data.
Can you actually see the devices and their data (other than the perf data)? I believe that the 32/64 bit issue was only with rrd data and that your ZODB database ought to be portable. The other technique I have used with several customers is to use the zenbatchdump / zenbatchload facilities to move devices to a new system. Use zenbatchdump -h for usage. Basically you can export your ZODB database, either en masse, or in bits - I often use device class as a parameter to move a bit at a time. It produces a text file which you can then inspect, edit, and prune devices that you don't want to import.
Cheers,
Jane
------------------------------
Jane Curry
Skills 1st United Kingdom
jane.curry@skills-1st.co.uk
------------------------------
Subject: |
RE: Zenoss 4.2.5 auto deploy script feedback |
Author: |
Clayton Tavernier |
Posted: |
2018-01-19 18:16 |
Ok, for example, right now if I look in Devices, I have a sub-category of Printers and in that I show 5 printers. If I click on one of them, I get the completely different webpage (not the error popup) that says "A Zenoss error has occurred" and offers to send the details of my error to Zenoss, Inc. If I "View Error Details" all it says is "getProductionState".
If I click on a device and try to remove it, I get the error popup that says "Unable to connect to server" that flashes by so fast that I can barely read it. If I try to remove several devices, I get the same error but the popup waits for me to click Reload. I get the same if I try to remove an entire sub-category.
Right now, all I want is my devices with their details moved over from my old server to my new server. I don't care about any history.
Is there a way to clear out all the device data on my new server and start from scratch? Or should I blow away the VM and start over?
------------------------------
Clayton
Network Computer Technician
------------------------------