One of our ES units has used 100% of its disk space and has only about 50 MB free and dropping every minute. I've purged as much data as I am willing but this had no apparent effect on disk space. Any ideas on how to check the file system to see what else could be taking up space?
ES is running out of Disk Space - Help!
Our staff is off this week so support is limited. Generally disk space is used by the /opt/mango/databases/mangoNoSQL folder for historical data or from a large number of events in the h2 database.
The other thing you can check is the /opt/mango/backup folder and delete extra backups. It’s possible incremental backups is turned on under the NoSQL Information in system settings.
You can go to the system setting page in mango and it should tell you the sizes but could take awhile if they are large. The log files are set to roll so probably not the issue.`
You can send a screen shot of the system settings if you want. If it does run out of space 100% its harder to recover from.
du -m /opt | sort -nr | head -n 20
On your MangoES should show you where the space is being used (you can change the path from /opt to drill up or down). As Joel says, it's possible that /opt/mango/backup directory is quite large.
You may also be able to recover some MB with
sudo apt-get clean
the default setting on the nosql database type is a purge resolution of 26 days. Changing the data purge settings by less than 26 days won't free up any space. The purge is executed in the middle of the night so you also have to wait a day to see if it made any difference.
If you upgrade to the latest nosql module there is a new setting for purging within shards which, if you enable it, will make the nosql module abide by the purge settings with the required resolution.
Good suggestions Craig, I'll just add that on the system setting page under the Purge section there is a button to run the purge immediately so you don't have to wait until the night.
There are a number of incremental NoSQL backup files but before I start deleting some are the original ones not needed to restore future ones?
In the root of databases dir there are three mah2* files of which the largest file is only 331 MB. Is this where most of the historical data is stored? Given that there are only 113 points with 4 months history and nothing more frequently stored than every minute I would think that this is all the data. We should therefore have no problem cleaning up a few gigs.
I'm not really sure what your first question is here? The backup zip files are only needed if you want to restore your database from a backup.
The actual NoSQL database (Historical Data) is stored in the databases/mangoNoSQL/ folder. 331 seems pretty large for the H2 Database and I'm guessing you have a lot of events that could be purged still.
Even with 113 data points if you are logging frequently it could amount to a lot of data. I would check your logging settings and make sure analog points that change a lot are set to something like a 1 minute average log and set points and binary data points for valves and pumps are on a log on change.