Product

Platform

Operating system

Error

Symptoms

Cause

Solution

Purpose of the DBCS_ArchiveIOError table:

- Logs an entry for each archive, retrieve, medium verify, medium logical copy

- Lists statistics on tapes and drives, including temporary errors

- Lists if error occurs on a particular tape or drive

- List what area of the tape a HRE was encountered in

- Tracks the throughput of each archive

How it works:

An entry will be logged into the dbcs_archiveioerror for each archiver operation that gets performed (archive, retrieve, verify ...). Medium Logical Copies consist of a number of different operations, so several entries will be logged into the table.

In addition, if a permanent read or write error is encountered on a medium, an entry will be logged specifically for that event.

IOERRORMEDMID:

The medium ID (GUID) of the tape on which the error occurs. This entry can be linked back to ARCHIVEMEDMID in the dbcs_archivemedium table. For example, to get all the records for the tape with name AGD546 SELECT * from dbcs_archiveioerror WHERE IOERRORMEDMID in (select ARCHIVEMEDMID from dbcs_archivemedium where archivemedmname='AGD546');

IOERRORTIME:

The time at which the error occurs - use the standard techniques for setting date formats if you want more/less precision.

IOERRORMEDMLOC:

The location indicates the extent of the operation or error for permanent errors. For instance, for a save-set operation it describes the current save-set. For a medium verify, it should describe the amount of the medium we verified. For a permanent read error from medium verify, the location indicates the start block and the number of blocks which are in a single contiguous unreadable chunk of the tape.

The medium location record is:

007 is a record written by Prinergy 2.0 (either an MTF or Araxi format tape).

003 is the Prinergy 2.0 format for Araxi 1.0 tapes.

<start block of saveset or error>,

<offset>, offset of start of item in save-set in logical blocks - only non-zero for a file in a saveset (for example, never in dbcs_archiveioerror).

<logical block count>, number of logical blocks occupied by item (saveset, error, file).

<logical block size>, actual physical block size for Araxi 1.0, 0 for medium verify meaning logical block count is in physical blocks.

For exampl, 007 337, 0, 448, 1024 is a record describing an operation on a saveset that starts at block 337 and spans 448 logical blocks. The logical block size here is 1024. This means that you are looking at an MTF tape. Araxi 1.0 format tape will display a physical blocksize value here (usually 16k). For example, 007 49333, 0, 5318, 16384.

IOERRORKINDCD:

This is a numeric code for the type of error record being written.

0 PERMANENT WRITE ERROR,

1 TEMPORARY (RECOVERED) WRITE ERROR,

2 PERMANENT READ ERROR,

3 TEMPORARY (RECOVERED) READ ERROR

IOERRORPHYSICALARCHIVEID:

The physical archive ID (GUID) of the archive where the error occurs ( for retrieve or archive operations). In the case of a verify, a value of !!!!!!!!!!!!!!!!!!!! will be recorded in this field.

IOERRORCORRECTEDERRORS:

The number of temporary (corrected) errors encountered.

IOERRORRETRIES:

The number of retries that were performed trying to complete this operation.

IOERRORDATATRANSFERRED:

The amount of data transferred during this operation. This is important because getting 4 temporary errors in 10GB is different from 4 temp errors in 100k. This value is recorded in MB as a floating point number.

IOERRORDATARATE:

The effective data transfer rate for the operation. This value is recorded in MB/s as a floating point number.

IOERRORDRIVEID:

The description and serial number of the physical drive in use when the error occurs.

IOERRORPERMANENTERRORS:

The number of permanent errors encountered.

IOERRORCORRECTEDERRORS:

The number of temporary (corrected) errors encountered.

IOERRORRETRIES:

The number of retries that were performed trying to complete this operation.

IOERRORDATATRANSFERRED:

The amount of data transferred during this operation. This is important because getting 4 temporary errors in 10GB is different from 4 temp errors in 100k. This value is recorded in MB as a floating point number.

IOERRORDATARATE:

The effective data transfer rate for the operation. This value is recorded in MB/s as a floating point number.

IOERRORDRIVEID:

The description and serial number of the physical drive in use when the error occurs.

Permanent fix:

See also

Keywords

archiver FAQs

Technical contact(s)

Amy Chan (CSC)

Submitted by

Amy Chan (CSC)

Date submitted

16 May 2003