Memory Info calculated wring for RHEL 7 Device
Hello all,
I have a Red Head Enterprise Linux 7 Device for which the Free Memory is calculated wrong. (I think) All worked fine until 3 Weeks ago, now I get a wrong free memory value.
The poller Module for Memopools reads: +-----------------+--------------------+-------+--------+--------+--------+--------+ | Label | Type | Index | Total | Used | Free | Perc | +-----------------+--------------------+-------+--------+--------+--------+--------+ | Virtual memory | HOST-RESOURCES-MIB | 3 | 47.5GB | 15.2GB | 32.3GB | 31.99% | | Memory buffers | HOST-RESOURCES-MIB | 6 | 15.5GB | 52.4MB | 15.4GB | 0.33% | | Swap space | HOST-RESOURCES-MIB | 10 | 32GB | 602MB | 31.4GB | 1.84% | | Physical Memory | UCD-SNMP-MIB | 0 | 15.5GB | 14.6GB | 912MB | 94.25% | +-----------------+--------------------+-------+--------+--------+--------+--------+
A free -m reads: total used free shared buff/cache available Mem: 15871 2629 884 8662 12357 4209 Swap: 32767 602 32165
Kind regards Mit freundlichen Grüßen Ing. Andreas Oberreiter (EDV) -- --- ÖBAU - Nadlinger --- Tel.: 02742/72042-79 Fax.: 02742/72042-7879 mailto:a.oberreiter@nadlinger.at www.nadlinger.at
Impressum: Baumarkt Nadlinger Handelsges.m.b.H. Porschestraße 29 3100 St. Pölten Firmenbuch FN 89076 w, LG St. Pölten DVR: 835935 Tel.: +43 2742 72042 Fax.: +43 2742 72042 7800 E-Mail.: mailto:nadlinger@nadlinger.at
I'm seeing this issue as well after updating to 19.10.10148, all my memory graphs went from low values to 90%+ mempool_perc. Is there a fix? For now, I updated my memory alert checks to be mempool_perc > 100 to disable the false alarms.
Thanks,
Graeme
On Mon, Oct 28, 2019 at 12:20 PM Oberreiter Andreas via observium < observium@observium.org> wrote:
Hello all,
I have a Red Head Enterprise Linux 7 Device for which the Free Memory is calculated wrong. (I think) All worked fine until 3 Weeks ago, now I get a wrong free memory value.
The poller Module for Memopools reads:
+-----------------+--------------------+-------+--------+--------+--------+--------+ | Label | Type | Index | Total | Used | Free | Perc |
+-----------------+--------------------+-------+--------+--------+--------+--------+ | Virtual memory | HOST-RESOURCES-MIB | 3 | 47.5GB | 15.2GB | 32.3GB | 31.99% | | Memory buffers | HOST-RESOURCES-MIB | 6 | 15.5GB | 52.4MB | 15.4GB | 0.33% | | Swap space | HOST-RESOURCES-MIB | 10 | 32GB | 602MB | 31.4GB | 1.84% | | Physical Memory | UCD-SNMP-MIB | 0 | 15.5GB | 14.6GB | 912MB | 94.25% |
+-----------------+--------------------+-------+--------+--------+--------+--------+
A free -m reads: total used free shared buff/cache available Mem: 15871 2629 884 8662 12357 4209 Swap: 32767 602 32165
Kind regards Mit freundlichen Grüßen Ing. Andreas Oberreiter (EDV) -- --- ÖBAU - Nadlinger --- Tel.: 02742/72042-79 Fax.: 02742/72042-7879 mailto:a.oberreiter@nadlinger.at www.nadlinger.at
Impressum: Baumarkt Nadlinger Handelsges.m.b.H. Porschestraße 29 3100 St. Pölten Firmenbuch FN 89076 w, LG St. Pölten DVR: 835935 Tel.: +43 2742 72042 Fax.: +43 2742 72042 7800 E-Mail.: mailto:nadlinger@nadlinger.at
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
This was caused by a fix for RHEL 8 having different memory reporting style to RHEL 7.
It seems the fix was a bit too enthusiastic, as predicted.
If everyone stopped using RHEL, we wouldn't have these issues, though...
Adam.
Sent from BlueMail
On 31 Oct 2019, 13:16, at 13:16, Graeme Davis via observium observium@observium.org wrote:
I'm seeing this issue as well after updating to 19.10.10148, all my memory graphs went from low values to 90%+ mempool_perc. Is there a fix? For now, I updated my memory alert checks to be mempool_perc > 100 to disable the false alarms.
Thanks,
Graeme
On Mon, Oct 28, 2019 at 12:20 PM Oberreiter Andreas via observium < observium@observium.org> wrote:
Hello all,
I have a Red Head Enterprise Linux 7 Device for which the Free Memory
is
calculated wrong. (I think) All worked fine until 3 Weeks ago, now I get a wrong free memory
value.
The poller Module for Memopools reads:
+-----------------+--------------------+-------+--------+--------+--------+--------+
| Label | Type | Index | Total | Used |
Free
| Perc |
+-----------------+--------------------+-------+--------+--------+--------+--------+
| Virtual memory | HOST-RESOURCES-MIB | 3 | 47.5GB | 15.2GB |
32.3GB
| 31.99% | | Memory buffers | HOST-RESOURCES-MIB | 6 | 15.5GB | 52.4MB |
15.4GB
| 0.33% | | Swap space | HOST-RESOURCES-MIB | 10 | 32GB | 602MB |
31.4GB
| 1.84% | | Physical Memory | UCD-SNMP-MIB | 0 | 15.5GB | 14.6GB |
912MB
| 94.25% |
+-----------------+--------------------+-------+--------+--------+--------+--------+
A free -m reads: total used free shared buff/cache available Mem: 15871 2629 884 8662 12357 4209 Swap: 32767 602 32165
Kind regards Mit freundlichen Grüßen Ing. Andreas Oberreiter (EDV) -- --- ÖBAU - Nadlinger --- Tel.: 02742/72042-79 Fax.: 02742/72042-7879 mailto:a.oberreiter@nadlinger.at www.nadlinger.at
Impressum: Baumarkt Nadlinger Handelsges.m.b.H. Porschestraße 29 3100 St. Pölten Firmenbuch FN 89076 w, LG St. Pölten DVR: 835935 Tel.: +43 2742 72042 Fax.: +43 2742 72042 7800 E-Mail.: mailto:nadlinger@nadlinger.at
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Most enterprises use RHEL/CentOS because of the long, stable lifecycle compared to other distros...
but, it indeed looks like there was a net-snmp bug -- I just updated net-snmp-5.7.2-43-el7 and Observium is happy! So I'm assuming there was Observium code to fudge with the incorrect data returned from older EL7 net-snmp versions?... ugh :(
https://bugzilla.redhat.com/show_bug.cgi?id=1601012 https://bugzilla.redhat.com/show_bug.cgi?id=1250060 https://access.redhat.com/errata/RHBA-2019:2239
Thanks,
Graeme
On Thu, Oct 31, 2019 at 9:35 AM Adam Armstrong adama@memetic.org wrote:
This was caused by a fix for RHEL 8 having different memory reporting style to RHEL 7.
It seems the fix was a bit too enthusiastic, as predicted.
If everyone stopped using RHEL, we wouldn't have these issues, though...
Adam.
Sent from BlueMail http://www.bluemail.me/r?b=15726 On 31 Oct 2019, at 13:16, Graeme Davis via observium < observium@observium.org> wrote:
I'm seeing this issue as well after updating to 19.10.10148, all my memory graphs went from low values to 90%+ mempool_perc. Is there a fix? For now, I updated my memory alert checks to be mempool_perc > 100 to disable the false alarms.
Thanks,
Graeme
On Mon, Oct 28, 2019 at 12:20 PM Oberreiter Andreas via observium < observium@observium.org> wrote:
Hello all,
I have a Red Head Enterprise Linux 7 Device for which the Free Memory is calculated wrong. (I think) All worked fine until 3 Weeks ago, now I get a wrong free memory value.
The poller Module for Memopools reads:
+-----------------+--------------------+-------+--------+--------+--------+--------+ | Label | Type | Index | Total | Used | Free | Perc |
+-----------------+--------------------+-------+--------+--------+--------+--------+ | Virtual memory | HOST-RESOURCES-MIB | 3 | 47.5GB | 15.2GB | 32.3GB | 31.99% | | Memory buffers | HOST-RESOURCES-MIB | 6 | 15.5GB | 52.4MB | 15.4GB | 0.33% | | Swap space | HOST-RESOURCES-MIB | 10 | 32GB | 602MB | 31.4GB | 1.84% | | Physical Memory | UCD-SNMP-MIB | 0 | 15.5GB | 14.6GB | 912MB | 94.25% |
+-----------------+--------------------+-------+--------+--------+--------+--------+
A free -m reads: total used free shared buff/cache available Mem: 15871 2629 884 8662 12357 4209 Swap: 32767 602 32165
Kind regards Mit freundlichen Grüßen Ing. Andreas Oberreiter (EDV) -- --- ÖBAU - Nadlinger --- Tel.: 02742/72042-79 Fax.: 02742/72042-7879 mailto:a.oberreiter@nadlinger.at www.nadlinger.at
Impressum: Baumarkt Nadlinger Handelsges.m.b.H. Porschestraße 29 3100 St. Pölten Firmenbuch FN 89076 w, LG St. Pölten DVR: 835935 Tel.: +43 2742 72042 Fax.: +43 2742 72042 7800 E-Mail.: mailto:nadlinger@nadlinger.at
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Yeah, that's the thing.
This is caused by RHEL doing exactly the opposite of what everyone thinks RHEL is for. They made a major unilateral change to an API that it's practically impossible to work out if you're talking to a changed version or not.
It's not even a software version thing, they modified the behaviour of net-snmp using their own patches, because who knows why.
But you know, people are still going to use them because "stable API" or some other such nonsense.
Adam.
Sent from BlueMail
On 31 Oct 2019, 17:09, at 17:09, Graeme Davis via observium observium@observium.org wrote:
Most enterprises use RHEL/CentOS because of the long, stable lifecycle compared to other distros...
but, it indeed looks like there was a net-snmp bug -- I just updated net-snmp-5.7.2-43-el7 and Observium is happy! So I'm assuming there was Observium code to fudge with the incorrect data returned from older EL7 net-snmp versions?... ugh :(
https://bugzilla.redhat.com/show_bug.cgi?id=1601012 https://bugzilla.redhat.com/show_bug.cgi?id=1250060 https://access.redhat.com/errata/RHBA-2019:2239
Thanks,
Graeme
On Thu, Oct 31, 2019 at 9:35 AM Adam Armstrong adama@memetic.org wrote:
This was caused by a fix for RHEL 8 having different memory reporting style to RHEL 7.
It seems the fix was a bit too enthusiastic, as predicted.
If everyone stopped using RHEL, we wouldn't have these issues,
though...
Adam.
Sent from BlueMail http://www.bluemail.me/r?b=15726 On 31 Oct 2019, at 13:16, Graeme Davis via observium < observium@observium.org> wrote:
I'm seeing this issue as well after updating to 19.10.10148, all my memory graphs went from low values to 90%+ mempool_perc. Is there a
fix?
For now, I updated my memory alert checks to be mempool_perc > 100
to
disable the false alarms.
Thanks,
Graeme
On Mon, Oct 28, 2019 at 12:20 PM Oberreiter Andreas via observium < observium@observium.org> wrote:
Hello all,
I have a Red Head Enterprise Linux 7 Device for which the Free
Memory is
calculated wrong. (I think) All worked fine until 3 Weeks ago, now I get a wrong free memory
value.
The poller Module for Memopools reads:
+-----------------+--------------------+-------+--------+--------+--------+--------+
| Label | Type | Index | Total | Used |
Free
| Perc |
+-----------------+--------------------+-------+--------+--------+--------+--------+
| Virtual memory | HOST-RESOURCES-MIB | 3 | 47.5GB | 15.2GB | 32.3GB | 31.99% | | Memory buffers | HOST-RESOURCES-MIB | 6 | 15.5GB | 52.4MB | 15.4GB | 0.33% | | Swap space | HOST-RESOURCES-MIB | 10 | 32GB | 602MB | 31.4GB | 1.84% | | Physical Memory | UCD-SNMP-MIB | 0 | 15.5GB | 14.6GB | 912MB | 94.25% |
+-----------------+--------------------+-------+--------+--------+--------+--------+
A free -m reads: total used free shared buff/cache available Mem: 15871 2629 884 8662 12357 4209 Swap: 32767 602 32165
Kind regards Mit freundlichen Grüßen Ing. Andreas Oberreiter (EDV) -- --- ÖBAU - Nadlinger --- Tel.: 02742/72042-79 Fax.: 02742/72042-7879 mailto:a.oberreiter@nadlinger.at www.nadlinger.at
Impressum: Baumarkt Nadlinger Handelsges.m.b.H. Porschestraße 29 3100 St. Pölten Firmenbuch FN 89076 w, LG St. Pölten DVR: 835935 Tel.: +43 2742 72042 Fax.: +43 2742 72042 7800 E-Mail.: mailto:nadlinger@nadlinger.at
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
participants (3)
-
Adam Armstrong -
Graeme Davis -
Oberreiter Andreas