Cargando…
Towards push notifications in OCIS
<!--HTML-->To make clients pick up changes to a shared file the etag of all recipients root folders needs to be updated. The current implementation in OCIS jails shares into a `/Shares` folder to calculate a dynamic etag, based on all accepted shares. This multiplies the stat requests made to...
Autor principal: | |
---|---|
Lenguaje: | eng |
Publicado: |
2021
|
Materias: | |
Acceso en línea: | http://cds.cern.ch/record/2750473 |
_version_ | 1780969104488792064 |
---|---|
author | Dreyer, Jörn |
author_facet | Dreyer, Jörn |
author_sort | Dreyer, Jörn |
collection | CERN |
description | <!--HTML-->To make clients pick up changes to a shared file the etag of all recipients root folders needs to be updated. The current implementation in OCIS jails shares into a `/Shares` folder to calculate a dynamic etag, based on all accepted shares. This multiplies the stat requests made to the underlying storage system by the number of shares on every propfind. By letting the storage registry cache the root etag of every storage id we can reduce this number to one stat request per storage. This cache would not only allow the gateway to calculate the etag for any path based on the mount point of storages: it would allow sharing the stat cache for all clients, the ocis instance could dynamically adjust the cache timeout to react to system load, and storages providers can push etag changes to the storage registry to prepare for push notifications to clients. |
id | cern-2750473 |
institution | Organización Europea para la Investigación Nuclear |
language | eng |
publishDate | 2021 |
record_format | invenio |
spelling | cern-27504732022-11-02T22:26:04Zhttp://cds.cern.ch/record/2750473engDreyer, JörnTowards push notifications in OCISCS3 2021- Cloud Storage Synchronization and SharingHEP Computing<!--HTML-->To make clients pick up changes to a shared file the etag of all recipients root folders needs to be updated. The current implementation in OCIS jails shares into a `/Shares` folder to calculate a dynamic etag, based on all accepted shares. This multiplies the stat requests made to the underlying storage system by the number of shares on every propfind. By letting the storage registry cache the root etag of every storage id we can reduce this number to one stat request per storage. This cache would not only allow the gateway to calculate the etag for any path based on the mount point of storages: it would allow sharing the stat cache for all clients, the ocis instance could dynamically adjust the cache timeout to react to system load, and storages providers can push etag changes to the storage registry to prepare for push notifications to clients.oai:cds.cern.ch:27504732021 |
spellingShingle | HEP Computing Dreyer, Jörn Towards push notifications in OCIS |
title | Towards push notifications in OCIS |
title_full | Towards push notifications in OCIS |
title_fullStr | Towards push notifications in OCIS |
title_full_unstemmed | Towards push notifications in OCIS |
title_short | Towards push notifications in OCIS |
title_sort | towards push notifications in ocis |
topic | HEP Computing |
url | http://cds.cern.ch/record/2750473 |
work_keys_str_mv | AT dreyerjorn towardspushnotificationsinocis AT dreyerjorn cs32021cloudstoragesynchronizationandsharing |