Discussion:
[systemd memory-leak] leak in systemd version 195
(too old to reply)
Shekhar arya
2017-11-23 16:30:35 UTC
Permalink
Raw Message
Hi Guys,

I am using systemd v195. With normal system operation, there is a little
increase in the system's memory. But, when I run certain commands to get
big files over ssh every 30 minutes, I see that the system go unresponsive
after 3 days. Further analysis led to the root cause of the issue as
systemd memory leak. There is an increase of around ~200MB in the VM. This
increase is contributed by ~44MB in heap and ~160MB in certain section of
the systemd which I am not sure.

1. Yellow section is the increase from 1.3MB to 45.6MB,


​

2. Again yellow section is the increase from 3.6MB to 168 MB. I am unsure
about this section. Can you guys help me understand this section which is
leaking memory?


​
--
regards,
- Shekhar
Reindl Harald
2017-11-23 16:36:44 UTC
Permalink
Raw Message
Post by Shekhar arya
I am using systemd v195
didn't you find any older version?
even CentOS7 has v219
Shekhar arya
2017-11-23 17:09:31 UTC
Permalink
Raw Message
My system uses poky 1.6 and it uses systemd v195. Since, it is stable till
now, we have not upgraded the systemd,
Post by Reindl Harald
Post by Shekhar arya
I am using systemd v195
didn't you find any older version?
even CentOS7 has v219
_______________________________________________
systemd-devel mailing list
https://lists.freedesktop.org/mailman/listinfo/systemd-devel
--
regards,
- Shekhar
Mantas Mikulėnas
2017-11-23 16:39:07 UTC
Permalink
Raw Message
Post by Shekhar arya
Hi Guys,
I am using systemd v195. With normal system operation, there is a little
increase in the system's memory. But, when I run certain commands to get
big files over ssh every 30 minutes, I see that the system go unresponsive
after 3 days. Further analysis led to the root cause of the issue as
systemd memory leak. There is an increase of around ~200MB in the VM. This
increase is contributed by ~44MB in heap and ~160MB in certain section of
the systemd which I am not sure.
v195 is five years old, already on the far end of "unsupported". It's
likely that the leak has been fixed long ago. If you absolutely cannot
upgrade, see if you can find something relevant in post-v195 commit logs.

My only guess is that each connection results in a new unit (maybe a
.scope, or an instance of ***@.service, or... whatever existed back then)
and those are not properly garbage-collected later.
--
Mantas Mikulėnas <***@gmail.com>
Sent from my phone
Shekhar arya
2017-11-23 17:10:02 UTC
Permalink
Raw Message
Thanks. I will check it.
Post by Mantas Mikulėnas
Post by Shekhar arya
Hi Guys,
I am using systemd v195. With normal system operation, there is a little
increase in the system's memory. But, when I run certain commands to get
big files over ssh every 30 minutes, I see that the system go unresponsive
after 3 days. Further analysis led to the root cause of the issue as
systemd memory leak. There is an increase of around ~200MB in the VM. This
increase is contributed by ~44MB in heap and ~160MB in certain section of
the systemd which I am not sure.
v195 is five years old, already on the far end of "unsupported". It's
likely that the leak has been fixed long ago. If you absolutely cannot
upgrade, see if you can find something relevant in post-v195 commit logs.
My only guess is that each connection results in a new unit (maybe a
then) and those are not properly garbage-collected later.
--
Sent from my phone
--
regards,
- Shekhar
Loading...