cloudfoundry-community/bolo-boshrelease

A BOSH release for deploying a Bolo Monitoring System

bolo-boshrelease's Issues

Perl collectors fail

They need some CPAN modules (include these in the bolo-collectors package) and a proper PERL5LIB

dbolo job can't colo with diego / garden-linux

This is the same busybox segfault we've seen before. Drop the helpers altogether on this release, since they don't get us much.

Turning on RRDCached fails

Tried to create a new deployment of Bolo this morning, and set cached: true on the rrd and gnossis jobs, to no avail. The graph images were corrupt, and the data wasn't being collected (although valid RRD files were being built).

I think this is a wiring problem with where the rrdcached socket file (UNIX domain socket) gets put, vs. where everything else is told to look for it.

Turning off caching cached: false "fixed" the problem, to the detriment of disk IO performance...

README.md for director uuid update is incorrect

Please change the README.md to update the director.yml instead of directory.yml

Custom RRA definitions

Update the rrd job to allow for custom RRAs, probably in a format that is easier than the raw RRDtool commands that bolo2rrd wants.

Enable RRDCached

The rrd job should (optionally?) make use of rrdcached for high-volume deployments.

New default graphs

Add default graphs for all of the process-specific checks:

• openfiles
• memory usage
• cpu usage
• io

Collector Autodetection

Add a new property, dbolo.autoconfigure, defaulting to true, which will cause the startup script to attempt to configure collectors automatically, while still honoring the explicit config in the manifest.

1. Enable the linux collector every 20s
2. Enable the process collector for all processes under monit
3. Check for the presence of known jobs and/or bound processes, and enable stuff like nginx and postgres collectors accordingly.

Graphs with (.*-)?<ITEM>-.* are not showing up

After deploying this to bosh-lite graps like this one that contain <ITEM> in the name are not being displayed in the gnossis front end. Only graphs without <ITEM> are visible.

Add marco/polo to BOSH release

https://github.com/bolo/marco

Need a marco job for handling the server side and the configuration.

Need a polo job for restarting dbolo in response to config changes from marco.

Data flow doesn't show on bosh-lite

Second row, left side.

clicking on bolo link requires un + pw

https://bolo.niftylogic.com/ (linked in readme) leads to a password protected page

Add bolo2meta everywhere it makes sense

Need to add a bolo2meta job, like the rrd job, for getting metadata about what the aggregator core is actually processing.

rrd graph error on new deployment

Error message:
ERROR: width below 10 pixels

Steps I did to get this error:

1. deployed new bolo-boshrelease via Genesis template to an aws account
2. visited the web UI
3. observed the graphs were all partially rendered

Log details here: https://gist.github.com/wayneeseguin/3ce081134f0fe27e1bf212ad0ebf3857

I also tried tweaking the URI parameters to the defaults for rrd graphs, no luck.