Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
INTRODUCTION ============ cron2rss is a simple set of tools designed to make your cron job (ie. scheduled background tasks) output into an RSS feed. RSS feeds can be tracked with your favourite RSS reader such as Firefox. But cron already knows how to email you the results of your background processes, so what's the point? Well, the result of a cron job can be one of: 1) The cron job ran and succeeded. 2) The cron job ran and failed. 3) The cron job didn't run. 4) The cron job started and is still running. Come back later. 5) The cron job started, then cron or the whole machine crashed. Cron emails are not very good for #1, because if you have a job running every 10 minutes, you don't want the email. Thus, almost everybody ensures that their cron jobs don't generate any output when they're successful, and thus they won't generate an email. You have to be careful when doing this, of course; if you just redirect all the stdout and stderr of your cron jobs to /dev/null, then you won't get emails in case #2 either - and those emails might have been really useful in that case! Also, case #3 is often overlooked but is a frequent source of problems. What if you cron daemon dies? What if the jobs are running on a server, and the server goes down altogether? You won't get an email in that case, and the *lack* of email is supposed to tell you there's a problem. But if you, like almost everyone, disabled log output for *successful* jobs, then the *normal* case is you don't get an email. That means you have no way of detecting when your server goes down, and that's not much good. cron2rss deals with your cron monitoring problems in these ways: 1) It keeps log output around even for successful runs. That means you don't have to redirect anything to /dev/null. cron2rss detects failed jobs by their exit code, not by their output. 2) It has an easy way of viewing *only* failed entries (ie. a separate RSS feed with only the failures) 3) It detects the *lack* of new log output for an unusually long time, and reports *that* as an error. 4) It detects tmpfiles made at the start of a job and not yet completed. 5) It makes it easy for you to monitor your logs from a host *other* than the ones running the jobs, so that if your job host goes down, the monitor host can still run. Plus, a single RSS feed can monitor all your cron jobs. 6) It's easy to set up multiple cron2rss monitor servers, so you can subscribe to the RSS feed from more than one machine in case cron2rss itself fails. HOW IT WORKS ============ Despite its usefulness, cron2rss operates very simply; a script called 'cron2rss/add' captures the exit code and stdout/stderr of your cron tasks and stores them in a set of log files. Each different cron task gets its output stored in a separate directory. Then, a script called 'cron2rss/expire' runs periodically to trim out old log files, plus create warning messages if any of the log directories hasn't received new input for a long time. If your cron jobs don't run very often, you can also run a script called 'cron2rss/stamp' more frequently. Its job is to make note that cron itself is running, even though no new log messages have appeared. This allows cron2rss to give you an early warning if it looks like your cron job eventually *won't* run, because it notices that the stamp file is out of date. Finally, a script called 'cron2rss/index.cgi' goes through the log messages, sorts them in reverse chronological order, and gives you an RSS feed with the results. Because its data structure is so simple (a simple tree of files under the cron2rss/data/ folder), it's also simple to replicate cron2rss data from one machine to another, and merge the logs from multiple machines into one. You can use the 'cron2rss/data-to' and 'cron2rss/data-from' scripts for this. BASIC INSTALLATION ================== You need: - a Unix system with cron and perl installed. - a web server (such as apache) set to serve .cgi files using ExecCgi. - a copy of the cron2rss files in a folder called 'cron2rss'. - an installed copy of rsync, if you want to sync the log files. Let's set up two simple cron tasks, as follows: a. Try to log into a remote server occasionally to make sure it's okay. b. Try to request a web page occasionally to make sure it works. Here's what we do. 1. Move your cron2rss files to your ~/public_html directory and cd there: mv cron2rss ~/public_html/cron2rss cd ~/public_html/cron2rss 2. Make data folders for your different tasks: mkdir data/task-a data/task-b 3. Start the cron job editor: crontab -e 4. Add lines for the cron2rss housekeeping tasks: 1 * * * * ~/public_html/cron2rss/expire 2 * * * * ~/public_html/cron2rss/stamp task-a task-b 5. Add lines for your actual tasks: 11 * * * * ~/public_html/cron2rss/add task-a ssh myhost true 0 4,8,12,16,20 * * * ~/public_html/cron2rss/add task-b wget -O/dev/null http://myserver.com/ 6. Save and exit the cron job editor. 7. Visit the cron2rss RSS feeds using your RSS-enabled web browser, such as Firefox: http://localhost/~username/cron2rss/ http://localhost/~username/cron2rss/failed/ Of course, replace 'localhost' with the name of the server where cron2rss is installed, and 'username' with your username where the public_html folder exists. The first URL shows *all* log entries, and the second one shows only *failed* log entries. Use the first one for now, as it's easier. 8. Make a couple of test log entries rather than waiting for cron: ~/public_html/cron2rss/add task-a true ~/public_html/cron2rss/add task-a ssh myhost true 9. Ensure that the entries show up in your RSS feed. If you go this far, then congratulations! cron2rss is working fine! MONITORING MULTIPLE SERVERS IN A SINGLE FEED ============================================ Let's say you have a second server, bobserver, that also has some tasks it needs to run, and you want to monitor those tasks too. Plus, bobserver tends to crash a lot, so you want to be sure to notice if you stop getting updates from it. Here's how you set it up, assuming you've already done the above steps. 1. Go to the cron2rss folder on your main server. cd ~/public_html/cron2rss 2. Install the cron2rss files on bobserver. ./install-to bobserver:~/cron2rss 3. Go over to bobserver and create a data folder for task-c. ssh bobserver cd cron2rss mkdir data/task-c 4. Start the cron job editor on bobserver: crontab -e 5. Add lines for the cron2rss housekeeping tasks: 1 * * * * ~/public_html/cron2rss/expire 2 * * * * ~/public_html/cron2rss/stamp task-c 6. Add lines for your actual task: 42 * * * * ~/public_html/cron2rss/add task-c df 7. Add a line to send the data back to your main server: 45 * * * * ~/public_html/cron2rss/data-to mainserver:public_html/cron2rss 8. Save and exit the cron job editor. Now, the directories for task-a and task-b will be created and updated on mainserver, and task-c will be sent to mainserver from bobserver. All three tasks can be viewed using the RSS feed. If bobserver crashes, then it won't be able to transmit any new data into the task-c folder on mainserver. Thus, the cron2rss/expire task on mainserver will eventually notice that task-c hasn't been updated, and issue its own warning. You'll then find out that bobserver isn't working right. Perfect! There's one more catch: what if, like most people, your random servers (such as bobserver) aren't supposed to be able to ssh into your log server (mainserver)? Well, you can do it the other way too. On mainserver, set up a cron entry like this: 45 * * * * ~/public_html/cron2rss/data-from bobserver:cron2rss Note that we're using data-from here instead of data-to. Good luck! Avery email@example.com