Download statistics are not updated for GlassFish project
Summary
The main download page of the Eclipse GlassFish project uses the Find a mirror script to download GlassFish downloadables. But the Eclipse Download Stats page doesn't find any statistics for most of the downloadables.
Steps to reproduce
- Go to the GlassFish download page - https://glassfish.org/download.html
- Click to download the latest GlassFish version, currently "Eclipse GlassFish 7.0.10, Jakarta EE Platform, 10". This will use the Find a mirror script via the URL https://www.eclipse.org/downloads/download.php?file=/ee4j/glassfish/glassfish-7.0.10.zip. This will correctly download the glassfish-7.0.10.zip file
- Logged in with an Eclipse account, go to the Eclipse Download Stats page https://dev.eclipse.org/committers/committertools/stats.php
- Search for "glassfish" in "download.glassfish.org", select "Last 30 days"
What is the current bug behavior?
The search in the Eclipse Download Stats page will yield results for "/glassfish/glassfish-5.1.0.zip" but no stats for "/ee4j/glassfish/glassfish-7.0.10.zip" or any other glassfish-7*.zip file.
What is the expected correct behavior?
The GlassFish download page uses the Find a mirror script to download GlassFish zip files for some time already. I would expect at least some statistics for glassfish-7.0.10.zip, or older versions like glassfish-7.0.9.zip, glassfish-7.0.8.zip, etc., not only statistics for glassfish-5.1.0.zip.
Relevant logs and/or screenshots
It looks like the glassfish-5.1.0.zip statistics is still being updated, there are download numbers for the last month. GlassFish 7 files, which are not displayed in the statistics search, are in a different directory, in '/ee4j/glassfish/', while the GlassFish files that are in the statistics report are in '/glassfish/' directory.
Priority
- Urgent
- High
- Medium
- Low
Severity
- Blocker
- Major
- Normal
- Low
Impact
The GlassFish team has no information about how many people download the released binaries and how popular GlassFish server is.
Activity
added IT-prioritylow IT-severitylow team:webdev labels
added statetodo label
This appears to be related to #78 (closed) , specifically the exclusion of *ee4j*(relevant as your file path is /ee4j/glassfish which matches) from mirroring, which also excludes it from stats tracking.
AIR that was done in part because the ee4j project was using some other tool to generate their stats, but it may have had a larger blast radius then expected.
Thanks @mward for your reply.
Is it really true that download statistics are not gathered if mirroring is disabled? Can the mirror script be adjusted to count statistics for all resources and only disable mirroring for excluded resources?
We don't need mirroring for Eclipse GlassFish artifacts, but we would still like to gather statistics from downloads. That's why we use the mirror script for the downloads, only to gather statistics about downloads, no need to use mirrors.
I actually do think those GlassFish artifacts should be mirrored. But the exception rule implemented in #78 (closed) excluded all Eclipse EE4J artifacts from being mirrored. Now that I looked at it in more detail, there's this comment: #78 (comment 318715), which says: "/ee4j: nothing of importance, or stuff that is already gone." This is simply not true - the ee4j directory contains GlassFish releases, and maybe it contains releases of other projects in the EE4J top level project.
Ideally, GlassFish directory
/ee4j/glassfish/should be included in mirroring. I don't know why the wholeee4jdirectory was excluded, but at least the glassfish subdirectory contains relevant downloadables that should be mirrored. The GlassFish team is also interested in download statistics, which is another reason to enable mirroring for that subdirectory.If memory serves me correctly, it was excluded as links were pointing to download.eclipse.org directly, bypassing the mirrors. There was no point in replicating that content.
As you're using the mirror page for downloads, I will remove /ee4j from the exclusion list. Your users will get closer downloads, you will get stats, we will save bandwidth. Everyone wins.
assigned to @droy
FWIW, my comment about "nothing of importance" was related to the directory contents at the time (which, as I recall, was not much, as the project was using download.e.o/glassfish, as I recall). It was in no way related to the output or content produced by the project. Apologies if that comment seemed insensitive.
The comment was fine. I understand we're all busy people. And, at that time, the ee4j was probably really not used. The GlassFish project started to use the mirroring script much later. It's all due to the fact that the whole EE4J space is huge and it's taking a long time until all those new projects get into shape. I apologize, if I sounded too harsh, I didn't mean it
