I’m still having problems with the new file-system in 6.4. After a power failure some programs (apache server for example) doesn’t start properly because “corruped file system detected” is detected for example in logs files…
It doesn’t happen with mysql files, but still fails with others (apache)…
If you power down without first shutting Apache down, it may leave log files in an unacceptable state. The solution is to run chkfsys at startup. In the worst case you might have to “zap” some log files. MySQL uses a special procedure to protect against an inconsistent database during an unscheduled power loss. It does this at a significant lost of speed, but then it is your data it is protecting.
Thanks for feedback!.. As I mentioned in other topic we run chkfsys always as standard start-up configuration from boot image (with unmounted partition) in order to repair the FS. That is what we are doing in QNX 6.3… My (wrong) asumption was that the new QNX6 FS would not need to perform chkfsys. But is not the case.
Everything shows me that QNX6 FS still needs always to run chkfsys at startup sequence.
So, the problem is much worse!.. If chkfsys does not repair the FS, who does?. If there are open files when the power goes down, it MUST be a secure method to restart with no problem… so I really hope chkfsys repairs the FS in this cases, if not, we are in troubles… aren’t we?
If its fs-qnx4.so then hopefully chkfsys will do the job. If its fs-qnx6.so try chkqnx6fs. But there is what chkqnx6fs doc says:
You shouldn’t actually need to use chkqnx6fs in a production system (e.g. in a boot script). The design of the fs-qnx6 filesystem should (in the absence of software bugs, physical bad blocks, or malicious data modification on the raw device) make any such check unnecessary.
I had already read those notes about fs-qnx6.so, and I have still many doubts the benefits of it, in (our) real world. You’ve seen the hardware we use, nothing strange. You said I’ve to go deeper on the hardware, but actually I have no hope it helps…
From the docs:
The Power-Safe filesystem was designed for and is intended for traditional rotating hard disk drive media.
That’s all what we have!. Maybe I have to investigate some file system options…
If that were the case, there should be no problem.
If the file system goes from one safe state to another safe state, I could lose some data (data in cache)… but NO to have the disk in an unsecure state (so some procceses doesn’t start ok after a power-loss)…
So, what is the real problem? Why my hard disk information is not well consolidated?
According to docs, my device has to support FLUSH/SYNC command. How can I know this ?