antrea
antrea copied to clipboard
antrea-io
Run Multiple Kind CI Jobs on a single testbed
1.Estimated size required for a single job run is ~4gb, after each job run success/failure, custom image must be cleaned to allow other jobs to start. ( It is standard practice ) 2.If space is not available, kind jobs will be queued. Need to get more data by running multiple kind job preferably before merge. 3.With the change we can maintain fewer kind test beds and get to know min/max no of kind test beds required to be maintained.
1.Estimated size required for a single job run is ~4gb, after each job run success/failure, custom image must be cleaned to allow other jobs to start. ( It is standard practice ) 2.If space is not available, kind jobs will be queued. Need to get more data by running multiple kind job preferably before merge. 3.With the change we can maintain fewer kind test beds and get to know min/max no of kind test beds required to be maintained.
We can't say that the estimated size for the job is 4gb, it is actually double the size, because we will have multiple images getting downloaded for test, then we will have kubernetes cluster, and different containers running as well. When I mentioned size 4GB, it was the size that was occupied on my local docker desktop, by the antrea images.
1.Estimated size required for a single job run is ~4gb, after each job run success/failure, custom image must be cleaned to allow other jobs to start. ( It is standard practice ) 2.If space is not available, kind jobs will be queued. Need to get more data by running multiple kind job preferably before merge. 3.With the change we can maintain fewer kind test beds and get to know min/max no of kind test beds required to be maintained.
We can't say that the estimated size for the job is 4gb, it is actually double the size, because we will have multiple images getting downloaded for test, then we will have kubernetes cluster, and different containers running as well. When I mentioned size 4GB, it was the size that was occupied on my local docker desktop, by the antrea images.
For 2 & 3, need to have actual space identified for a single job to decide on no of kind test beds.
1.Estimated size required for a single job run is ~4gb, after each job run success/failure, custom image must be cleaned to allow other jobs to start. ( It is standard practice ) 2.If space is not available, kind jobs will be queued. Need to get more data by running multiple kind job preferably before merge. 3.With the change we can maintain fewer kind test beds and get to know min/max no of kind test beds required to be maintained.
We can't say that the estimated size for the job is 4gb, it is actually double the size, because we will have multiple images getting downloaded for test, then we will have kubernetes cluster, and different containers running as well. When I mentioned size 4GB, it was the size that was occupied on my local docker desktop, by the antrea images.
For 2 & 3, need to have actual space identified for a single job to decide on no of kind test beds.
but if we will have more testbed, then what will be the purpose of multiple kind jobs on single testbed, we can either leverage one testbed with more resources.
1.Estimated size required for a single job run is ~4gb, after each job run success/failure, custom image must be cleaned to allow other jobs to start. ( It is standard practice ) 2.If space is not available, kind jobs will be queued. Need to get more data by running multiple kind job preferably before merge. 3.With the change we can maintain fewer kind test beds and get to know min/max no of kind test beds required to be maintained.
We can't say that the estimated size for the job is 4gb, it is actually double the size, because we will have multiple images getting downloaded for test, then we will have kubernetes cluster, and different containers running as well. When I mentioned size 4GB, it was the size that was occupied on my local docker desktop, by the antrea images.
For 2 & 3, need to have actual space identified for a single job to decide on no of kind test beds.
but if we will have more testbed, then what will be the purpose of multiple kind jobs on single testbed, we can either leverage one testbed with more resources.
It will depend on max allowed space we can reserve for a kind vm, even if run ci jobs in parallel , we might not be allowed to reserve a powerful vm which can run all ci jobs in parallel , however benefit is that we would reduce no of vms needed for running kind ci tests.
LGTM, I will wait for @KMAnju-2021 's confirmation regarding the jenkins kind verification before approving. Please ensure that all related Jenkins jobs can pass.
@KMAnju-2021 , need to change executor in jenkins for kind testbed and then multiple kind jobs can run a single kind test bed, then trigger multiple kind jobs here
/test-kind-conformance /test-kind-ipv6-only-conformance /test-kind-ipv6-conformance /test-kind-ipv6-only-networkpolicy /test-kind-ipv6-only-e2e /test-kind-ipv6-e2e /test-kind-networkpolicy /test-kind-ipv6-networkpolicy
/test-kind-conformance /test-kind-ipv6-only-conformance /test-kind-ipv6-conformance
/test-kind-ipv6-only-networkpolicy /test-kind-networkpolicy /test-kind-ipv6-networkpolicy
LGTM, I will wait for @KMAnju-2021 's confirmation regarding the jenkins kind verification before approving. Please ensure that all related Jenkins jobs can pass
@XinShuYang kind tests are passed..
LGTM, for kind CI changes, @KMAnju-2021 do you have more comments?
Hi @XinShuYang, I don't have any comments, thanks !
I don't see any status for test-kind-networkpolicy and test-kind-e2e. Is there an issue?
I don't see any status for
test-kind-networkpolicyandtest-kind-e2e. Is there an issue?
@KMAnju-2021 can you check the trigger comments.
I don't see any status for
test-kind-networkpolicyandtest-kind-e2e. Is there an issue?
Hi @antoninbas, fixed test-kind-networkpolicy , and test-kind-e2e trigger phrase is not valid, thanks!