Hello,

I a building a product where I use a SD media on an external USB SD reader formatted with the ext3 file system with full journaling enabled under CentOS 4.3. Everything works okay but occasionally (usually anywhere from 6000 loops to 40000 loops) I get a very long stall when either setting the fstream file pointer using the seekp method or doing a write. In this situation the code enters the seekp or write call and does not return until 34 to 35 seconds. It seems like it is timing out internally somehow.

I have tried changing the /sys/block/sda/device/timeout value from its default of 30 to 1 but this does not seem to change anything. Note that fstream.bad() is not set when the call returns. Is there anything else I can do to lessen the amount of time the call stalls in seekp or write? Does anybody know why this could or would happen? I have tried this with multiple readers and lots of different media always with the same results.

Additional info: The device is mounted synchronously, data journaling is enabled and commit time is set to 1 second.

Any help or suggestions would be greatly appreciated

Moved: This thread is more suitable in Programming and has been moved accordingly to help your thread/question get the exposure it deserves.

After doing more research I have found the problems were due to media. It turns out that when using SD media if you want high quality with no errors you really must go high grade.

With the high quality media I don't get those long timeouts anymore. Since the software is the same and the reader is the same, I assume it has to be some type of I/O timeout. It would be great if I could reduce the length of those timeouts but I can live with it for now with the high quality media.