HELP HELP Messed up my complete Cacti installation

[geertn444]

Hi all,

I need some advanced help recovering my Cacti installation.
I have 9000 graphs with two items: traffic_in and traffic_out. I was playing aroung with the graph template (including removing the data input field) and now all my graphs fail.

I know the cause: all my graphs miss the second parameter (see first attachement).
As you can see, i need to (manually) lookup up the corresponding traffic_out of the same interface and put this in, after that it is fixed.

However, i want to do this directly in the SQL database, because i need to fix +/- 5000 of these graphs !

so my question is: where can i find the link between the graph_id and the two input parameters ?

The system is still polling the values into the data sets, just the link between the graph and the traffic_out dataset is lost.

Please help

[geertn444]

We need to find out the ID of the RRD file used in the traffic_in direction
I don't know in which table i can find the unique id of an RRD. is this the id field in the data_template_id table ?

i discovered that when i lookup an id in this table, then pick the "local_data_id" in the same row and query the table "data_template_data_rra" for this id, i get two rows: traffic_in and traffic_out contained inside this RRD.

where this is all linked to graph's remains unclear.

[BSOD2600]

sounds like the simple solution would be to rollback to your last good weekly database backup. Don't overwrite the rrd files though.

[geertn444]

Can you help me how to do this ?

is there a simple "rollback" command.

I don't know if there are regular backups of the database.

[BSOD2600]

I don't know if there are regular backups of the database.

Those would be something you would have set up... like all good system adminstrators do.

[rony]

If you are lucky, you just altered the template. If you fix the template, all the broken data sources should work again.

[TheWitness]

Per Rony's comment, do a fresh install of Cacti in another location, or your desktop. Export the Data Query in question and import it to your production system.

Then, in the future, schedule a cronjob that does something like the following:

Code: Select all

#!/bin/bash
# My Backup Script
BACKUPDIR=<some_location>
NUMBER=10

# See how many backups I have
BACKUPS=`ls -1 $BACKUPDIR | wc -l | tr -d ' '`

# Backup the Database
mysqldump cacti > $BACKUPDIR/cacti_backup$$.sql

# Remove old Backups
if [ $BACKUPS -ge $NUMBER ]; then
  NUMBERTOREMOVE=`expr $BACKUPS - $NUMBER`
  FILES=`ls -altr $BACKUPDIR | head -$NUMBERTOREMOVE`
  for FILE in $FILES; do
    /bin/rm -f $FILE
  done
fi

Schedule this daily in Cron, that way you will have better restoration options. Or simply run this before you go playing around. Then getting backto a good point will be as easy as:

Code: Select all

mysql cacti < mylastbackup.sql

TheWitness