LizardFS Room for Improvement
Well, if you don't have a support contract and therefore don't have acess to the automated failover mechanism, you need to build it yourself. This can be achieved by using common open source software, but you will need to know what you are doing. It can be a painful process since you will need to write some scripts on your own (e.g. an OCF agent if you decide to use Pacemaker and CRM).
View full review »Management console - today it is managed via scripts and flat files which need to be synchronized among all master/shadow nodes in a cluster. There are severall possible problems in such approach:
- it is possible to 'damage' config by mistake
- it is required to copy modified config file to other servers (typically you change it on master server and populate this to other - shadow servers using scp). If one forgets this, shadow servers have incorrect/old config and in case the cluster failovers the new master service is using wrong config
I would suggest to write a dedicated program or script which checks syntax and do all the changes in a proper way in background. It could look like this:
to create new export:
- lizard-admin create-export
to modify:
- lizard-admin modify-export
etc.
Buyer's Guide
Software Defined Storage (SDS)
March 2024
Find out what your peers are saying about LizardFS, Red Hat, Nutanix and others in Software Defined Storage (SDS). Updated: March 2024.
768,740 professionals have used our research since 2012.