URL: https://github.com/SSSD/sssd/pull/1018 Author: alexey-tikhonov Title: #1018: mem-cache: sizes of free and data tables were made consistent Action: opened
PR body: """ This is partial backport of #999
Since size of "free table" didn't account for SSS_AVG_*_PAYLOAD factor only small fraction of "data table" was actually used. SSS_AVG_*_PAYLOAD differentiation for different payload types only affected size of hash table and was removed as unjustified.
Resolves: https://pagure.io/SSSD/sssd/issue/4160 """
To pull the PR as Git branch: git remote add ghsssd https://github.com/SSSD/sssd git fetch ghsssd pull/1018/head:pr1018 git checkout pr1018
URL: https://github.com/SSSD/sssd/pull/1018 Author: alexey-tikhonov Title: #1018: mem-cache: sizes of free and data tables were made consistent Action: synchronized
To pull the PR as Git branch: git remote add ghsssd https://github.com/SSSD/sssd git fetch ghsssd pull/1018/head:pr1018 git checkout pr1018
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
Label: +Waiting for review
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
sumit-bose commented: """ Hi,
thank you for the patch, but I have to admit that I would prefer to have a more minimal patch for 1-16 which only fixes the too small free table issue. E.g. something like `mc_ctx->ft_size = MC_FT_SIZE(mc_ctx->dt_size);` (not tested).
bye, Sumit """
See the full comment at https://github.com/SSSD/sssd/pull/1018#issuecomment-610194402
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
alexey-tikhonov commented: """
but I have to admit that I would prefer to have a more minimal patch for 1-16 which only fixes the too small free table issue. E.g. something like `mc_ctx->ft_size = MC_FT_SIZE(mc_ctx->dt_size);` (not tested).
`ft_size` should be based on amount of slots (not bytes) it data table.
But most importantly, could you please explain what is the functional difference / what sense does it make to stick to misleading `SSS_AVG_*_PAYLOAD`?
I think there is (almost) no change from user perspective / observed behavior. But code is simpler/cleaner and thus easier to maintain. """
See the full comment at https://github.com/SSSD/sssd/pull/1018#issuecomment-610228026
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
alexey-tikhonov commented: """
but I have to admit that I would prefer to have a more minimal patch for 1-16 which only fixes the too small free table issue. E.g. something like `mc_ctx->ft_size = MC_FT_SIZE(mc_ctx->dt_size);` (not tested).
`ft_size` should be based on amount of slots (not bytes) it data table.
But most importantly, could you please explain what is the functional difference / what sense does it make to stick to misleading `SSS_AVG_*_PAYLOAD`?
I think there is (almost) no change from user perspective / observed behavior. But code is simpler/cleaner and thus easier to maintain.
I also think it makes sense for "core" components to keep difference between 1-16 and master branch small (where possible). """
See the full comment at https://github.com/SSSD/sssd/pull/1018#issuecomment-610228026
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
sumit-bose commented: """
but I have to admit that I would prefer to have a more minimal patch for 1-16 which only fixes the too small free table issue. E.g. something like `mc_ctx->ft_size = MC_FT_SIZE(mc_ctx->dt_size);` (not tested).
`ft_size` should be based on amount of slots (not bytes) it data table.
ah, sorry, I forgot `/MC_SLOT_SIZE` which should be safe here but is nevertheless ugly.
But most importantly, could you please explain what is the functional difference / what sense does it make to stick to misleading `SSS_AVG_*_PAYLOAD`?
I think there is (almost) no change from user perspective / observed behavior. But code is simpler/cleaner and thus easier to maintain.
I also think it makes sense for "core" components to keep difference between 1-16 and master branch small (where possible).
My thinking was to keep the difference in the 1-16 branch as small a possible to make it more easy to spot changes which caused unexpected behavior or issues.
But you are right, since it is expected that we will backport patches to 1-16 for some time, it might be easier to keep the difference to master small.
I'll run some tests with the current version.
bye, Sumit
"""
See the full comment at https://github.com/SSSD/sssd/pull/1018#issuecomment-610330570
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
Label: -Waiting for review
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
Label: +Accepted
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
pbrezina commented: """ * `sssd-1-16` * 38d4eabf1a8e61b4facdd3f2cb385498471d1ea9 - mem-cache: sizes of free and data tables were made consistent
"""
See the full comment at https://github.com/SSSD/sssd/pull/1018#issuecomment-611469974
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
Label: +Pushed
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
Label: -Accepted
URL: https://github.com/SSSD/sssd/pull/1018 Title: #1018: mem-cache: sizes of free and data tables were made consistent
Label: -Ready to push
URL: https://github.com/SSSD/sssd/pull/1018 Author: alexey-tikhonov Title: #1018: mem-cache: sizes of free and data tables were made consistent Action: closed
To pull the PR as Git branch: git remote add ghsssd https://github.com/SSSD/sssd git fetch ghsssd pull/1018/head:pr1018 git checkout pr1018
sssd-devel@lists.fedorahosted.org