I’ve been playing around with a virtualized clone of our production Graylog system to test upgrading us from Graylog 2.3 and Elasticsearch 2.3 to Graylog 3.1 with Elasticsearch 5.6+.
I notice when I upgraded ES from 2.3 to 5.6 that it scrambled all my index folders and file names. Previously my indexes were all each in it’s own directory like so:
graylog2_1040 graylog2_1073 graylog2_1106 graylog2_1139 graylog2_1172 graylog2_1205 graylog2_1238 graylog2_1271
After the upgrade they are each in a directories that look like this:
akLcyuXvSNOVFRWbilfppA h0Z7taXWSNydzKl_w0_kDg mFc63z4_ReyQ0ONzuRwimw tCsrTBT_Skq0zAQFd6saqw Zwk9yx0BTMC_cGLp3MpxDw
AnwGoarRR7K2bzA88Gd5ew H7UtruDPRGCQYGAGpFkGwQ MHfyQr-MT72nt-JJrtA3Iw
So instead of the nice friendly “graylog2_1040” I see folders and files coded like “MHfyQr-MT72nt-JJrtA3Iw” which are hard on the eyes. I realize this is more of an ES question then Graylog, but:
How can I prevent ES from doing this again when I do this same upgrade again in production?