[ClusterLabs] some comments on corosync.conf(5)

Ulrich Windl Ulrich.Windl at rz.uni-regensburg.de
Tue Apr 23 06:25:38 EDT 2019


Hi!

Reading the corosync.conf manual page of corosync 2.3.6 (SLES12 SP4), I have some random comments:

The tag indent for "clear_node_high_bit" seems broken.

Shouldn't "hold" be "token_hold"? Why is "token_retransmits_before_loss_const" so long (why the "_const")?
The description states "...retransmit and hold will be automatically  calculated...". Shouldn't that be "token_retransmit" (also see my earlier remark on "hold")?

For "token" the page says "Reforming a new configuration takes about 50 milliseconds..." shouldn't "50 ms" be "join timeout"?

"consensus should be larger then token" : ".. larger than .."?
"... which stil guarantee...": "still guarantee"? Maybe try a spell-checker ;-)

In general the manual page would benefit if all parameter names were set in bold face, as in "as node count grows if consensus is less than token".

Terrible sentences: "This  constant specifies how many rotations of the token without receiving any of the messages when messages should  be  received may occur before a new configuration is formed."

"This specifies the number of times a problem is detected ...": "...a problem has to be detected..."?

"If this value is set, retransmit and hold will be automatically  calculated  from retransmits_before_loss and token.": the parameter "retransmits_before_loss" is not documented!



Regards,
Ulrich







More information about the Users mailing list