You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
readahead=4092
readahead=>4092
works just fine as expected.
Things get unnecessary complicated, error prone and buggy when one tries to specify the value in sectors instead of kib:
readahead=4092 s
readahead = 4092 s
-> The sector value is simply divided by 2 and passed the same way as the kib value.
IMO this redundancy is not needed and should not be allowed.
In this case dividing by 2 results in a float type which gets parsed to a string when applied to sysfs and one might get:
2024-11-06 09:52:03,273 ERROR tuned.plugins.base: verify: failed: device md127: 'readahead' = '2048', expected '2048.0'
readahead=>4092 s
Here things seem to break even worse:
2024-11-19 07:51:42,945 WARNING tuned.plugins.base: cannot compare new value '8192 s' with current value '512' by operator '>', using '>8192 s' directly as new value
For the "=>" operator it seems to be a bad idea to pass such non integer or double meaning types.
This part is a bit more abstract and it nicely shows that such (redundant/unnecessary) complexity is a bad idea. People should just use kib?
I can come up with a patch if this is considered the correct approach.
The text was updated successfully, but these errors were encountered:
readahead=4092
readahead=>4092
works just fine as expected.
Things get unnecessary complicated, error prone and buggy when one tries to specify the value in sectors instead of kib:
readahead=4092 s
readahead = 4092 s
-> The sector value is simply divided by 2 and passed the same way as the kib value.
IMO this redundancy is not needed and should not be allowed.
In this case dividing by 2 results in a float type which gets parsed to a string when applied to sysfs and one might get:
2024-11-06 09:52:03,273 ERROR tuned.plugins.base: verify: failed: device md127: 'readahead' = '2048', expected '2048.0'
readahead=>4092 s
Here things seem to break even worse:
2024-11-19 07:51:42,945 WARNING tuned.plugins.base: cannot compare new value '8192 s' with current value '512' by operator '>', using '>8192 s' directly as new value
For the "=>" operator it seems to be a bad idea to pass such non integer or double meaning types.
This part is a bit more abstract and it nicely shows that such (redundant/unnecessary) complexity is a bad idea. People should just use kib?
I can come up with a patch if this is considered the correct approach.
The text was updated successfully, but these errors were encountered: