Releases: pganalyze/collector
Releases · pganalyze/collector
v0.9.5
- Improved first user experience
- Add "--reload" command for sending SIGHUP to daemon process
- Show error message when configuration file is empty
- Experimental build support for Solaris
- System metrics: Various fixes
- Packaging: Add support for Ubuntu Precise / 12.04 LTS
v0.9.4
- Introduce "pganalyze-collector-helper": Setuid Binary that can be used to run
privileged actions when the main collector is running as non-root (the default) - Determine the correct distance between two collector runs (instead of assuming 600 seconds)
- Better monitoring for self-hosted systems
- Collect missing Disk I/O statistics
- Fix calculation logic for disk utilization
- Collect kernel version and architecture
- Don't monitor the local loopback network interface
- Sort disk/partition/network interface names before output
- Don't collect local system information when monitoring remote hosts
- Packaging
- Update to Go 1.7
- Systemd: Enforce memory limit of 256mb for the collector
v0.9.3
v0.9.2
v0.9.1
v0.9.0
v0.9.0rc8
- Significant restructuring of the codebase
- We're now sending data using the protocol buffers format
- Snapshot data is directly uploaded to S3
- Query, table and system statistics are diff-ed on the client side
- Support for monitoring system metrics on self-hosted systems is added again
- New safety mechanisms against stuck/slow collector runs
v0.9.0rc7
v0.9.0rc6
v0.9.0rc5
- Introduce new --diff-statements option (default off for now)
- This calculates the diff for the counter values of pg_stat_statements on the client (i.e. collector),
instead of the server for increased accuracy and protection against out-of-order processing
- This calculates the diff for the counter values of pg_stat_statements on the client (i.e. collector),
- Introduce "opts" to the snapshot, for indicating which options were chosen
- Never open more than 1 connection to the same database
- This covers edge cases like sending a lot of SIGHUP signals
- In case we detect more than 1 connection we error out and exit, to avoid
clogging the database
- Use AWS EC2 instance role if no credentials are specified