Fixed lambda_function to prevent errors during simultaneous deployments #48
Conversation
Signed-off-by: saimedhi <[email protected]>
saimedhi
requested review from
CEHENKLE,
dblock,
setiah,
kartg,
nknize,
tlfeng,
AMoo-Miki,
ashwin-pc and
saratvemulapalli
as code owners
| alias_status = os_client.indices.update_aliases( | ||
| { | ||
| "actions": [ | ||
| {"remove": {"index": old_index, "alias": INDEX_ALIAS}}, |
There was a problem hiding this comment.
Previously, we tried to remove old index. This will cause issues with parallel deployments. When we try to remove old index, it may have already been deleted by another parallel thread. So, here ideal solution will be to remove all the indices that are present and add new index. By this we can always have number of indices for a index alias as one or zero(at initial state).
There was a problem hiding this comment.
The information in these indices come from various sources. When parallel requests land, we do want both of them. The current code recognizes that it has lost the reference to the old index and decides to fail rather that have invalid data. By not removing the old ones, this PR would only mask the failure but result in invalid data.
What we need is for the process to recognize that there is another one running and wait for its completion before it continues.
Description
Fixed lambda_function to prevent errors during simultaneous deployments
Issues Resolved
closes #46
Check List
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.