Read: I/O error on disk - HP-UX E25 10.01

[Dee_3]

Received the Read:i/o error when trying to bkup or copy a file yet when I do a: dd if=/dev/rdsk/c0t5d0 of=dev/null bs=2k against both my hard drives to determine which one is bad, they both come back without errors. logtool not showing any errors either. My question - has anyone received read I/O errors on a file that prevented from copying and yet the disk was good? This is a first for me ---
help!

[A. Clay Stephenson · Accepted solution]

Yes, it is entirely possible for this situation to occur even if all the blocks on this LVOL or disk are perfectly fine. Consider this possibility, suppose that a file's metadata (inode, directory entry, ..) tells the OS to read block 104523 but there is no such block. The filesystem actually ends at block 92000, for example. The filesystem's metadata has asked for a read that is not possible. While there is no PHYSICAL problem with the device there is a LOGICAL problem.

Your next step is do do an fsck ( a full fsck if this is a vxfs filesystem).

[S.K. Chan]

Yes, matter of fact happened to me last week, I was migrating a particular user's home directory from one server to another. Did a remote "find-cpio" and it always hangs at one particular file. At first I thought it was just another case of disk gone/going back but after thorough checking (including running cstm's exerciser and fsck of the FS) it still does not go through for the second "find-cpio". This time I noticed it stop at the same file (a 1.2MB ProE file). It cpio'ed half of the file content and hang. So called the user, ask him to open the file with ProE and it seems fine. So what I did was just try a simply cp command of the file to local /tmp and it hung as well. Got fed up and asked the user if I could delete the file and he gave his blessing. So the third find-cpio finally went through ...
Just sharing my experience here ..
In your case you may want to run exerciser just to make sure.