crc32, unix time and other metrics should stay int so they can be compared, instead the output coes out as float. size in megabytes instead of bytes is less of an issue, but it would be nice to have it consistent.

Hi!

The docker hub image is still v0.3.5 whereas the v0.3.6 has already been released here.
Is there a possibility that you could update the docker hub image to v0.3.6 as well?

I was trying to build the dist-windows-386 but the package is removed immediatly and building just with dist only builds amd64 versions.

Hiya,

Appreciate the filestat exporter.

Would it be possible to add sparse file detection and its actual byte sizing?

Thanks
A

The 0.3.6 version starts to report an error

OS Version: CentOS Linux release 7.6.1810 (Core)

/usr/local/prometheus/filestat_exporter: /lib64/libc.so.6: version GLIBC_2.32' not found (required by /usr/local/prometheus/filestat_exporter) /usr/local/prometheus/filestat_exporter: /lib64/libc.so.6: version GLIBC_2.34' not found (required by /usr/local/prometheus/filestat_exporter)

[NEW Feature]
it's possible to add a scheduled monitoring feature to activate file monitoring at a specific time.

We want to be able to list all files matching a certain pattern in a directory or subdirectories of that directory. This is not possible in the current situation. Something like ** which works for Python globs would be a nice feature. For example when using "test/**/*.txt" as a pattern, this would look for all .txt files in the test directory and all its subdirectories.

Hello , encountered the following problem, there are 3 files test.txt here is my filestat.yaml

exporter:
  listen_address: ':9943'
  metrics_path: /metrics

  working_directory: "/app"
  # Default enable/disable of metrics - overridden if not set by parameter '-metric.*'
  enable_crc32_metric: true
  # enable_nb_line_metric: false
  # list of patterns to apply - metrics can be enable/disabled for each group
  files:
    - patterns: ["test1.txt","test2.txt","test23.txt"]
      enable_nb_line_metric: true
      enable_crc32_metric: true

The received metrics look like this and when changing the content of files hash change, but when deleting a file, for example test1.txt metrics do not change in any way. Also if you recreate the file test1.txt. the metrics will not change in any way.

HELP file_content_hash_crc32 CRC32 hash of file content using the IEEE polynomial
TYPE file_content_hash_crc32 gauge
file_content_hash_crc32{path="test1.txt"} 3.85946783e+09
file_content_hash_crc32{path="test2.txt"} 2.657284769e+09
file_content_hash_crc32{path="test23.txt"} 7.73236185e+08

I would be grateful to find a solution.

Is there any way to monitor files in different directories? For example

• /etc/passwd
• /data/blabla

Is there any way to monitor directories with this exporter??
I tried following patterns but it didn't work:

• .
• *
• **
• /**/

This is one of those edge case things that might not affect many instances. By the way this exporter is a lifesaver for a very specific problem I have been dealing with.

Problem: The physical order of the patterns in the config file seems to affect the count.

Example: yaml

...
  files:
  - patterns:
    -  "/main_dir/*"
    -  "/main_dir/Troubled_file*"
...

Result:
main_dir count will be accurate, but the specific file pattern within main_dir are grossly under counted. While ls -l /main_dir/Troubled_file* | wc -l would show around 500 of the specific files exporter would show ~0-1 sometimes 10.

Fix: yaml

...
  files:
  - patterns:
    -  "/main_dir/Troubled_file*"
    -  "/main_dir/*"
...

Result fix:
main_dir file count and specific file pattern within main_dir seem to be more accurate.

Probably not worth a code change, but maybe a note in the documentation.

Hi,

Appreciate the filestat exporter.

Would it be possible to get owner of the file?

Thanks
Ravi

Is there a way to monitor file backup availabe for last 24 hrs.
Since i was trying with

time() - file_stat_modif_time_seconds{instance="server:63387"} < 86400

• This expression will list the files which are avaliable in last 24 hours. But how can i alert which all files are not available in last 24 hrs.

It seems a final slash is missing on all the output:

filestat.yml:

exporter:
  #! Working directory of exporter
  working_directory: "/dir1/dir2/dir3/"

  #! Network parameters
  #listen_address: ':9943'
  #metrics_path: /metrics

  #! Uncomment one of the following to enable default config
  #enable_crc32_metric: true
  #enable_nb_line_metric: true

  # list of patterns to apply
  files:
    - patterns: ["*/*/.ERROR"]
    - patterns: ["*/*/*.ERROR"]
    - patterns: ["*/*/*/*.ERROR"]

Metric output:

file_glob_match_number{pattern="*/*/*.ERROR"} 294
file_glob_match_number{pattern="*/*/*/*.ERROR"} 36
file_glob_match_number{pattern="*/*/.ERROR"} 0
file_stat_size_bytes{path="dir/dir/dir20201208164827.xml.ERROR"} 171
file_stat_size_bytes{path="dir/dir/dir20201208165023.xml.ERROR"} 17

The files are actually located at:

/dir1/dir2/dir3/dir/dir/dir/20201208164827.xml.ERROR
/dir1/dir2/dir3/dir/dir/dir/20201208165023.xml.ERROR

Between the filename and the last directory a slash seems missing.
Is this expected, an issue with configuration, a bug or me not understanding something?

Thanks!

I'm using this exporter mostly to keep an eye on the number of files in a specific folder, and it works beautifully.

However, I don't need to have the file size of modified date to be recorded for every file that appears in a folder.

For folders that are used heavily this generates a huge amount of metrics generated that are never used and put a some strain on our Prometheus installation. It might be useful to have the ability to enable/disable specific metrics from being generated for use cases like this.

Nice exporter!
Is it possible to have shell expansions (command substitution) in the pattern of the config file?
I have tried but not got it to work.
If it isn't possible today, could you implement it?

For example to catch a file with todays date:

/tmp/*_`date +%Y%m%d`.log