Sorry, I missed title, send again.
Hello, Milosz,
Another question:
I see your documentation suggest to use pdudaemon. But I had a look for you validation lava lab: https://validation.linaro.org/scheduler/device/hi6220-hikey-r2-02/devicedict, I see you use /usr/local/lab-scripts/snmp_pdu_control
I want to know why your lab not use the one which you suggest? I had a look for pdudaemon, found it won't wait there until PDU operation finish. Will this have any issue? Will next logic start before the pdu really finish the operation?
On Mon, 18 May 2020 at 10:28, Peter Lucas peterlucas073@gmail.com wrote:
Sorry, I missed title, send again.
Hello, Milosz,
Another question:
I see your documentation suggest to use pdudaemon.
Could you point me to that. In the past we had a lot of issues with pdudaemon but IIRC these were fixed
But I had a look for you validation lava lab: https://validation.linaro.org/scheduler/device/hi6220-hikey-r2-02/devicedict, I see you use /usr/local/lab-scripts/snmp_pdu_control
https://git.linaro.org/lava/lava-lab.git/tree/shared/lab-scripts/snmp_pdu_co...
I want to know why your lab not use the one which you suggest? I had a look for pdudaemon, found it won't wait there until PDU operation finish. Will this have any issue? Will next logic start before the pdu really finish the operation?
I think issues with pdudaemon are now fixed. It's biggest drawback was that it implemented 'requests queue'. So jobs would often time out if the PDU operations were delayed by pdudaemon. IIRC pdudaemon implemets pass-through mode and acts as a library rather than a daemon, so this issue should be fixed. But you need to double check that. As you noticed we're not using pdudaemon in our lab. The tool to use strongly depends on the PDU you're using. The script I shared above works with APC PDUs that have SNMP enabled.
milosz
Lava-users mailing list Lava-users@lists.lavasoftware.org https://lists.lavasoftware.org/mailman/listinfo/lava-users
Thanks, Milosz,
What I saw in your doc is this: https://validation.linaro.org/static/docs/v2/pdudaemon.html Our PDU make also encounter issue when do concurrent operations, so if no special trap to use it in LAVA, we will have a try.
On Mon, May 18, 2020 at 7:15 PM Milosz Wasilewski < milosz.wasilewski@linaro.org> wrote:
On Mon, 18 May 2020 at 10:28, Peter Lucas peterlucas073@gmail.com wrote:
Sorry, I missed title, send again.
Hello, Milosz,
Another question:
I see your documentation suggest to use pdudaemon.
Could you point me to that. In the past we had a lot of issues with pdudaemon but IIRC these were fixed
But I had a look for you validation lava lab:
https://validation.linaro.org/scheduler/device/hi6220-hikey-r2-02/devicedict, I see you use /usr/local/lab-scripts/snmp_pdu_control
https://git.linaro.org/lava/lava-lab.git/tree/shared/lab-scripts/snmp_pdu_co...
I want to know why your lab not use the one which you suggest? I had a
look for pdudaemon, found it won't wait there until PDU operation finish. Will this have any issue? Will next logic start before the pdu really finish the operation?
I think issues with pdudaemon are now fixed. It's biggest drawback was that it implemented 'requests queue'. So jobs would often time out if the PDU operations were delayed by pdudaemon. IIRC pdudaemon implemets pass-through mode and acts as a library rather than a daemon, so this issue should be fixed. But you need to double check that. As you noticed we're not using pdudaemon in our lab. The tool to use strongly depends on the PDU you're using. The script I shared above works with APC PDUs that have SNMP enabled.
milosz
Lava-users mailing list Lava-users@lists.lavasoftware.org https://lists.lavasoftware.org/mailman/listinfo/lava-users
On Wed, 20 May 2020 at 03:05, Peter Lucas peterlucas073@gmail.com wrote:
Thanks, Milosz,
What I saw in your doc is this: https://validation.linaro.org/static/docs/v2/pdudaemon.html Our PDU make also encounter issue when do concurrent operations, so if no special trap to use it in LAVA, we will have a try.
The docs are a bit outdated. Both LAVA and PDUdaemon changed a lot since it was written. However I asked around and it looks like passthrough mode isn't yet implemented in pdudaemon. There is a PR submitted: https://github.com/pdudaemon/pdudaemon/pull/78 So to my best understanding at this point pdudaemon implements queuing which might cause trouble in case a lot of jobs starts at the same time.
milosz
On Mon, May 18, 2020 at 7:15 PM Milosz Wasilewski milosz.wasilewski@linaro.org wrote:
On Mon, 18 May 2020 at 10:28, Peter Lucas peterlucas073@gmail.com wrote:
Sorry, I missed title, send again.
Hello, Milosz,
Another question:
I see your documentation suggest to use pdudaemon.
Could you point me to that. In the past we had a lot of issues with pdudaemon but IIRC these were fixed
But I had a look for you validation lava lab: https://validation.linaro.org/scheduler/device/hi6220-hikey-r2-02/devicedict, I see you use /usr/local/lab-scripts/snmp_pdu_control
https://git.linaro.org/lava/lava-lab.git/tree/shared/lab-scripts/snmp_pdu_co...
I want to know why your lab not use the one which you suggest? I had a look for pdudaemon, found it won't wait there until PDU operation finish. Will this have any issue? Will next logic start before the pdu really finish the operation?
I think issues with pdudaemon are now fixed. It's biggest drawback was that it implemented 'requests queue'. So jobs would often time out if the PDU operations were delayed by pdudaemon. IIRC pdudaemon implemets pass-through mode and acts as a library rather than a daemon, so this issue should be fixed. But you need to double check that. As you noticed we're not using pdudaemon in our lab. The tool to use strongly depends on the PDU you're using. The script I shared above works with APC PDUs that have SNMP enabled.
milosz
Lava-users mailing list Lava-users@lists.lavasoftware.org https://lists.lavasoftware.org/mailman/listinfo/lava-users
lava-users@lists.lavasoftware.org
-
Milosz Wasilewski -
Peter Lucas