Abysmal performance of spool file staging from MoM->server on qrerun (and server->MoM when job runs again)

Description

Bad performance with large spool file when job is rerun.

1) First of all, MoM will use TCP PBS_BATCH_MvJobFile requests so move the files over to the server. Sadly, it will do so in extremely small requests that only contain a chunk of RT_BLK_SZ bytes from the file.

2) The server likewise uses TCP to send back the spool files to the new MoM when it runs a new runversion of the job. That, too, uses very small blocks (SCRIPT_CHUNK_Z) that each generate an interserver message with a PBS_BATCH_MvJobFile request.

Acceptance Criteria

None

Status

Assignee

Subhasis Bhattacharya

Reporter

Subhasis Bhattacharya

Severity

None

OS

None

Start Date

None

Pull Request URL

None

Story Points

1

Components

Fix versions

Affects versions

14.1.0

Priority

High
Configure