markow
/
ckpool
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

ckdb - update startup comments

master
kanoi 10 years ago
parent
b0d99751bf
commit
9891eff85b
1 changed files with 12 additions and 9 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 21
      src/ckdb.c

21
src/ckdb.c
Unescape View File

@ -84,8 +84,9 @@ static char *restorefrom;
* Thus we can queue all messages: * Thus we can queue all messages:
* workinfo, shares, shareerror, ageworkinfo, poolstats, userstats * workinfo, shares, shareerror, ageworkinfo, poolstats, userstats
* and block * and block
* to be processed after the reload completes and just process authorise * with an ok.queued reply to ckpool, to be processed after the reload
* messages immediately while the reload runs * completes and just process authorise messages immediately while the
* reload runs
* This can't cause a duplicate process of an authorise message since a * This can't cause a duplicate process of an authorise message since a
* reload will ignore any messages before the last DB auths message, * reload will ignore any messages before the last DB auths message,
* however, if ckdb and ckpool get out of sync due to ckpool starting * however, if ckdb and ckpool get out of sync due to ckpool starting
@ -106,16 +107,18 @@ static char *restorefrom;
* 1) the disk had not yet written it to the CCL when ckdb read EOF and * 1) the disk had not yet written it to the CCL when ckdb read EOF and
* ckpool was started at about the same time as the reload completed. * ckpool was started at about the same time as the reload completed.
* This can be seen if the message displayed in the fatal error IS NOT * This can be seen if the message displayed in the fatal error IS NOT
* in ckdb's message logfile. A ckdb restart will resolve this * in ckdb's message logfile.
* A ckdb restart will resolve this
* 2) ckpool was started at the time of the end of the reload, but the * 2) ckpool was started at the time of the end of the reload, but the
* authorise message was written to disk and found in the CCL before * message was written to disk and found in the CCL before it was
* it was processed in the message queue. This can be seen if the * processed in the message queue.
* message displayed in the fatal error IS in ckdb's message logfile * This can be seen if the message displayed in the fatal error IS in
* and means the messages after it in the logfile have already been * ckdb's message logfile and means the messages after it in the
* processed. Again, a ckdb restart will resolve this * logfile have already been processed.
* Again, a ckdb restart will resolve this
* In both the above (very rare) cases, if ckdb was to continue running, * In both the above (very rare) cases, if ckdb was to continue running,
* it would break the synchronisation and could cause DB problems, so * it would break the synchronisation and could cause DB problems, so
* ckdb aborting and needing a restart resolves this * ckdb aborting and needing a complete restart resolves it
* The users table, required for the authorise messages, is always updated * The users table, required for the authorise messages, is always updated
* immediately and is not affected by ckpool messages until we * immediately and is not affected by ckpool messages until we
* TODO: allow bitcoin addresses - this will also need to be handled * TODO: allow bitcoin addresses - this will also need to be handled

Write Preview
Loading…
Cancel
Save