Hi Dave,
Do you mean LAVA scheduler couldn't pick the device to consume the job? If scheduler couldn't pick device, it's too simple and strange, a job will run on any device, whether x86_64 or aarch64, Kunpeng or other server.
Thanks Yinsi
liuyinsi@163.com
From: Dave Pigott Date: 2021-04-30 18:11 To: liuyinsi@163.com CC: Luca Di Stefano; Anmar Oueja; wufengguang; Jonathan Cameron; lkq-dev; Chase Qi; jammy.zhou@linaro.org Subject: Re: Issues with docker on OpenEulerOS Hi Yinsi,
The trouble with this is we would have to change the test definitions for the Kungpen server - which would make life complicated for the CI front end because when a job is submitted LAVA scheduler just picks then first available device. You never know which worker it’s going to end up on.
Dave
On 30 Apr 2021, at 09:14, liuyinsi@163.com wrote:
Hi Dave,
Fix git clone failed: Please try to use http://gitee.com/liu-yinsi/test-definitions, it's a mirror of http://github.com/Linaro/test-definitions, and it will update automatically once a day.
Thanks Yinsi
liuyinsi@163.com
From: Dave Pigott Date: 2021-04-13 17:39 To: Jammy Zhou CC: liuyinsi; Luca Di Stefano; Anmar Oueja; Shameerali Kolothum Thodi; wufengguang; Guohanjun (Hanjun Guo); Jonathan Cameron; lkq-dev; Chase Qi Subject: Re: Issues with docker on OpenEulerOS Hi Jammy,
This is all good - not sure what, if anything, the Lab can do to fix these problems. This would appear to be a CI/Front end issue.
Thanks
Dave
On 13 Apr 2021, at 03:07, Jammy Zhou jammy.zhou@linaro.org wrote:
I checked with Chase who has good experience of LAVA in China. It looks like a cache/mirror server is helpful in this context.
The download speed from S3 often is a problem for China. We may need to setup a mirror/cache server (e.g. jfrog artifactory) in the lab to mirror or cache the files that LAVA test jobs need. Once we have a cache/mirror server set, the test plan should be updated to download file from mirror/cache server instead of the original S3 server. A local mirror/cache server will improve both reliability and speed of CI.
Regards, Jammy
On Mon, 12 Apr 2021 at 17:34, Dave Pigott dave.pigott@linaro.org wrote: Hi Yinsi,
I had to offline the worker over the weekend because the download speed was making jobs fail due to timeouts cloning and wgetting
Not much we as the lab team can do about that.
Dave
On 9 Apr 2021, at 15:19, liuyinsi@163.com wrote:
Hi All,
It's great to collaborate. :)
Thank you all.
Yinsi
liuyinsi 邮箱：liuyinsi@163.com 签名由 网易邮箱大师 定制 On 04/09/2021 19:52, Luca Di Stefano wrote: Hi All, Thank you Yinsi for all the support. Just as a note, to have a working setup for LAVA dispatcher through docker-compose both the backported kernel fix and the Qemu version from debian backports repo were needed in order to achieve this. Thank you all. Luca
On 09/04/2021 12:46, Dave Pigott wrote: Okay - great news. The Kungpen server is now online and the qemu devices are passing their health checks and now processing jobs.
Thank you all!
Dave
On 9 Apr 2021, at 04:57, liuyinsi@163.com wrote:
OK，done. Please try to login.
Thanks Yinsi
liuyinsi 邮箱：liuyinsi@163.com 签名由 网易邮箱大师 定制 On 04/08/2021 20:06, Luca Di Stefano wrote: Hi Yinsi, Other IP addresses that are relevant to the LAVA lab in addition to what Dave already said are: - 51.148.40.18 - 51.148.40.7 - 51.148.40.11 Thank you. Luca
On 08/04/2021 13:02, Dave Pigott wrote: Hi Yinsi,
The Lab IP is 51.148.40.1
Thanks
Dave
On 8 Apr 2021, at 12:30, liuyinsi@163.com wrote:
Hi Dave/Anmar,
The firewall is enabled for some reasons. Please send your public IP address to me, i need to add your public IP address to the trustlist.
Thanks Yinsi
liuyinsi 邮箱：liuyinsi@163.com 签名由 网易邮箱大师 定制 On 04/07/2021 18:21, Dave Pigott wrote: Hi Yinsi,
We can no longer ssh to the server. Can you resolve this please?
Thanks
Dave
On 7 Apr 2021, at 10:17, liuyinsi@163.com wrote:
Ok，done. Please try to test QEMU again.
Thanks Yinsi
On 04/01/2021 20:57, Anmar Oueja wrote: On Fri, Mar 19, 2021 at 12:26 PM Shameerali Kolothum Thodi shameerali.kolothum.thodi@huawei.com wrote: Ok, I have backported the fix to 5.10.
Please find the patches here, https://github.com/hisilicon/kernel-dev/commits/private-v5.10-GICv2-on-v3
Sanity testing at my end shows it working. Please try and let me know.
Thanks Shmeer.
On our side, we tried the newest stable version of QEMU and it didn't work unfortunately. The only way to fix this is either deploying the kernel above or a firmware fix.
Yinsi: Are you able to build and install the kernel in the above tree? Hanjun of Fengguang might be able to help.
anmar
From: Jonathan Cameron Sent: 19 March 2021 09:13 To: liuyinsi@163.com; jammy.zhou@linaro.org; Guohanjun (Hanjun Guo) guohanjun@huawei.com; Pigott dave.pigott@linaro.org Cc: Anmar Oueja anmar.oueja@linaro.org; Luca Di Stefano luca.distefano@linaro.org; wufengguang wufengguang@huawei.com; lkq-dev lkq-dev@op-lists.linaro.org; Shameerali Kolothum Thodi shameerali.kolothum.thodi@huawei.com Subject: RE: Re: Issues with docker on OpenEulerOS
We’ll still need the backport as I think the fix only went in during 5.11 but hopefully not to major. @Shameer, let us know if you hit any problems.
Thanks,
Jonathan
From: liuyinsi@163.com [mailto:liuyinsi@163.com] Sent: 19 March 2021 08:29 To: jammy.zhou@linaro.org; Jonathan Cameron jonathan.cameron@huawei.com; Guohanjun (Hanjun Guo) guohanjun@huawei.com; Pigott dave.pigott@linaro.org Cc: Anmar Oueja anmar.oueja@linaro.org; Luca Di Stefano luca.distefano@linaro.org; wufengguang wufengguang@huawei.com; lkq-dev lkq-dev@op-lists.linaro.org; Shameerali Kolothum Thodi shameerali.kolothum.thodi@huawei.com Subject: Re: Re: Issues with docker on OpenEulerOS
@Jammy,
Our dailybuild already make v5.10 for openEuler 21.03.
@Dave The openeuler OS host kernel has been updated to v5.10, because I have restarted the host, the previously running container has stopped.
Thanks Yinsi liuyinsi@163.com
From: Jammy Zhou Date: 2021-03-19 08:29 To: Jonathan Cameron; Guohanjun (Hanjun Guo) CC: Dave Pigott; liuyinsi; Anmar Oueja; Luca Di Stefano; wufengguang; lkq-dev; Shameerali Kolothum Thodi Subject: Re: Issues with docker on OpenEulerOS +Hanjun
As I know, the kernel will be based on v5.10 for openEuler 21.09, but for 21.03 and 20.03 LTS, it is still based on v4.19.
On Fri, 19 Mar 2021 at 00:33, Jonathan Cameron Jonathan.Cameron@huawei.com wrote: @yinsi
What are chances of upgrading the host os to a more recent openEuler kernel. 4.19 is a bit of a stretch for a backport of the fix. I guess it might be easy, but would feel more comfortable with 5.10 or 5.5
Or can we ask openEuler to carry a backport of this set directly?
Another alternative is to see if there is an available BIOS with the fix in place and upgrade.
Jonathan
On Thu, 18 Mar 2021 13:38:28 +0000 Dave Pigott dave.pigott@linaro.org wrote:
Hi Yinsi,
Okay - that fixed it. We’ve now got an excellent download speed. We’re back to the problems with qemu compatibility. Working on this.
Thanks for your help!
Dave
On 18 Mar 2021, at 12:11, liuyinsi@163.com wrote:
Hi Dave,
I pull the images in openEulerOS, though you got the message about docker pull limit, you still can pull successfully, ignore the message if not block our work.
Thanks Yinsi
liuyinsi 邮箱：liuyinsi@163.com https://maas.mail.163.com/dashi-web-extend/html/proSignature.html?ftlId=1&name=liuyinsi&uid=liuyinsi%40163.com&iconUrl=https%3A%2F%2Fmail-online.nosdn.127.net%2Fqiyelogo%2FdefaultAvatar.png&items=%5B%22%E9%82%AE%E7%AE%B1%EF%BC%9Aliuyinsi%40163.com%22%5D签名由 网易邮箱大师 https://mail.163.com/dashi/dlpro.html?from=mail88 定制
On 03/18/2021 19:47, Dave Pigott mailto:dave.pigott@linaro.org wrote: Hi Yinsi,
It’s lavasoftware/lava-dispatcher:2021.03
Dave
On 18 Mar 2021, at 11:44, liuyinsi@163.com mailto:liuyinsi@163.com wrote:
Hi Dave,
What is the docker images name you want to pull from dockerhub？
Thanks Yinsi
liuyinsi 邮箱：liuyinsi@163.com https://maas.mail.163.com/dashi-web-extend/html/proSignature.html?ftlId=1&name=liuyinsi&uid=liuyinsi%40163.com&iconUrl=https%3A%2F%2Fmail-online.nosdn.127.net%2Fqiyelogo%2FdefaultAvatar.png&items=%5B%22%E9%82%AE%E7%AE%B1%EF%BC%9Aliuyinsi%40163.com%22%5D签名由 网易邮箱大师 https://mail.163.com/dashi/dlpro.html?from=mail88 定制
On 03/18/2021 19:28, Dave Pigott mailto:dave.pigott@linaro.org wrote: Hi Yinsi,
We’re trying to upgrade the docker worker to the latest release and we got the message that we’ve hit our docker pull limit. We’ve only done two pulls. Any ideas?
Thanks
Dave
On 18 Mar 2021, at 10:47, liuyinsi@163.com mailto:liuyinsi@163.com wrote:
Hi Dave,
I notice the job 2313644 https://lkft.validation.linaro.org/scheduler/job/2313644 is created 2 weeks, 6 days ago, can you test a new job, test wget http://images.validation.linaro.org/snapshots.linaro.org/components/lava/sta...http://images.validation.linaro.org/snapshots.linaro.org/components/lava/standard/debian/sid/arm64/2/vmlinuz-4.6.0-1-arm64 on openEulerOS machine is ok, first download speed about 2.63MB/s.
It's better to improve your download code, to face unstable and slow network issues, for example, increase timeout and add retries.
Thanks Yinsi
liuyinsi@163.com mailto:liuyinsi@163.com From: Dave Pigott mailto:dave.pigott@linaro.org Date: 2021-03-18 17:04 To: liuyinsi@163.com mailto:liuyinsi@163.com CC: Anmar Oueja mailto:anmar.oueja@linaro.org; Luca Di Stefano mailto:luca.distefano@linaro.org; wufengguang mailto:wufengguang@huawei.com; Jonathan Cameron mailto:jonathan.cameron@huawei.com; lkq-dev mailto:lkq-dev@op-lists.linaro.org; jammy.zhou@linaro.org mailto:jammy.zhou@linaro.org Subject: Re: Issues with docker on OpenEulerOS Hi Yinsi,
If you look at a failed example - e.g. https://lkft.validation.linaro.org/scheduler/job/2313644https://lkft.validation.linaro.org/scheduler/job/2313644 - you’ll see it’s not failing on a git clone, it’s failing on a wget.
Dave
On 18 Mar 2021, at 08:19, liuyinsi@163.com mailto:liuyinsi@163.com wrote:
Hi Dave,
"http download timeout", does it happen often? "https download timeout", has this problem been solved by edit ~/.gitconfig?
Thanks Yinsi
liuyinsi@163.com mailto:liuyinsi@163.com From: liuyinsi@163.com mailto:liuyinsi@163.com Date: 2021-03-17 10:19 To: Dave Pigott mailto:dave.pigott@linaro.org CC: Oueja mailto:anmar.oueja@linaro.org; luca mailto:luca.distefano@linaro.org; wufengguang mailto:wufengguang@huawei.com; Cameron mailto:jonathan.cameron@huawei.com; lkq-dev mailto:lkq-dev@op-lists.linaro.org; jammy.zhou@linaro.org mailto:jammy.zhou@linaro.org Subject: Re: Re: Issues with docker on OpenEulerOS
> > Locally, in the lab, we use KissCache for https cacheing. We would have to go through every test definition submitted by every bot and developer to change from https to git. > vi ~/.gitconfig
[url "git://github.com git://github.com"] insteadOf = https://github.com https://github.com/
This will change from https to git.
Hi Yinsi,
If you look at the job definitions, e.g. https://lkft.validation.linaro.org/scheduler/job/2409008/definition#defline4...https://lkft.validation.linaro.org/scheduler/job/2409008/definition#defline40, you’ll see the URL is defined there. I’m not sure where you are suggesting this could be universally changed to use git.
Hi Dave,
You can edit the file in any environment where you execute the clone command, for example, i get the URL and will git clone in a qemu, if i edit ~/.gitconfig in qemu before clone, then it will automatically change https to git.
Thanks Yinsi
Dave
> > > > Thanks > > Dave > > >>>> >>>> >>>> anmar
liuyinsi 邮箱：liuyinsi@163.com 签名由 网易邮箱大师 定制