Add detail to initial Java dump topic #79
Conversation
|
Work in progress. @pshipton please can you review what I have so far for accuracy. I wrote it up based on a recent javacore but there are some explanations missing and possibly some errors. The core file did not have a SHARED CLASSES section as I guess it wasn't enabled. Can you provide me with something please? Still work to do for THREADS but that's all I have time for this week. |
SueChaplain
force-pushed
the
javadump
branch
from
1ed2c03 to
519d8cc
Compare
|
Attaching a javacore containing shared cache info. The cgroup info change is not delivered yet so the javacore does not contain this. |
| NULL | ||
| ``` | ||
|
|
||
| ## THREADS |
There was a problem hiding this comment.
There is no explanation of the THREADS section.
There was a problem hiding this comment.
Yep, WIP. Will get back to it soon. Thanks for the SHARED CLASSES output.
There was a problem hiding this comment.
SHARED CLASSES and THREADS sections now in place. I think that the topic now contains a relatively complete set of information. Do we know if the cgroup PR will go in to 0.10.0 please? If it will not, I'll comment out the parts that relate to it and I think this is good to merge.
What I'd like to do as a separate task is take 2 or 3 specific scenarios (hopefully from service) and create troubleshooting examples. E.g. I got a gpf, what can javadump tell me about the root cause? If we can supplement this topic with those examples, I think the info here is as good as it was in the J9 docs. If you have any thoughts on what the scenarios should be, please let me know.
There was a problem hiding this comment.
cgroup PR is still under review and blocked on the docs being ready. It doesn't need to go into 0.10.0, but it depends on when the review comments are suitably addressed and the branch for 0.10.0 occurs. In theory it should go in after 0.10.0 but no guarantees.
There was a problem hiding this comment.
gpf or assert, user signal, OOM, deadlock maybe.
There was a problem hiding this comment.
Thanks for the suggestions on scenarios.
For cgroup, I think the docs are ready .... I've got it in the release notes and what's new, plus this javadump topic contains a mention and example output. One of the developers is writing a blog about it, which I might be able to extract more information from later on. If I can get this topic merged I can close the doc issue. Let me know ....
There was a problem hiding this comment.
The cgroup review still needs to close before it can be merged, and the review comments haven't yet been addressed. We were hoping to branch for 0.10.0 tomorrow, but seems likely we won't be ready.
There was a problem hiding this comment.
cgroup information removed for 0.10.0
SueChaplain
force-pushed
the
javadump
branch
from
519d8cc to
2939e93
Compare
SueChaplain
force-pushed
the
javadump
branch
2 times, most recently
from
28b3209 to
e726783
Compare
docs/dump_javadump.md
Outdated
|
|
||
| In the following example, the shared classes cache was created with a Class Debug Area (`-Xnolinenumbers=false`). Byte code instrumentation (BCI) is enabled, which is the default, and VMs sharing the cache are allowed to store classpaths, which is also the default. | ||
|
|
||
| The `Cache Summary` shows a cache size (`2SCLTEXTCSZ`) of 16776608 bytes, with a soft maximum size (`2SCLTEXTSMB`) also of 16776608 bytes, which leaves 12691668 bytes of free space (`2SCLTEXTFRB`). The size of the Class Debug Area (`2SCLTEXTDAS`) is 1331200 bytes, which uses only 11% of the available space. |
There was a problem hiding this comment.
11% is the amount used inside of the class debug area, i.e. 156240 is used out of 1331200.
docs/dump_javadump.md
Outdated
| - `3XMTHREADINFO3`: Java callstack information (`4XESTACKTRACE`) or Native call stack information (`4XENATIVESTACK`). | ||
| - `5XESTACKTRACE`: This line indicates whether locks were taken by a specific method. | ||
|
|
||
| Java thread priorities are mapped to operating system priority values. The larger the thread priority, the more frequently the thread runs. |
There was a problem hiding this comment.
The larger the thread priority, the more frequently the thread runs.
Not necessarily true https://www.linuxjournal.com/node/10267
There was a problem hiding this comment.
OK, removed that statement.
SueChaplain
force-pushed
the
javadump
branch
2 times, most recently
from
46b5d55 to
84404d7
Compare
|
Jenkins doc stage |
Includes sections to cover all the different parts that make up a Java dump file, with explanations for the data. Makes use of a very recent javacore. Signed-off-by: Sue Chaplain <[email protected]>
SueChaplain
force-pushed
the
javadump
branch
from
84404d7 to
860009b
Compare
SueChaplain
changed the title
Includes sections to cover all the different
parts that make up a Java dump file, with explanations
for the data. Makes use of a very recent javacore
that includes the new cgroup information for
eclipse-openj9/openj9#1932
Signed-off-by: Sue Chaplain [email protected]