Looking at the docs:
|
For the asset and data object indices the Generic Data Index uses an alias (e.g. `<index_prefix>_asset`) that points to the |
|
most current index (e.g. `<index_prefix>_asset-odd`). The alias name always stays the same, the index names alternate |
|
between `-odd` and `-even` suffix. For more details also see 'Updating index structure for data indices' in the next section. |
it is not clear why Pimcore creates an even and an odd version of each index. Having 2x the number of indexes means 2x the number of shards. We are bumping up against resource constraints due to the large number of shares (2/data class) and wanted to (a) confirm / understand why there are two version of each and (b) see if there are any solutions to reduce the shard count.
Looking at the docs:
generic-data-index-bundle/doc/02_Configuration/03_Index_Management.md
Lines 26 to 28 in f76d347
it is not clear why Pimcore creates an even and an odd version of each index. Having 2x the number of indexes means 2x the number of shards. We are bumping up against resource constraints due to the large number of shares (2/data class) and wanted to (a) confirm / understand why there are two version of each and (b) see if there are any solutions to reduce the shard count.