Greenplumn failed to build index: No space left on device

when you look at the error prompt, it feels like there is not enough space, but it is not.
5 machines each have 20T disks, each with 8T left. But every time the index is created for about a day or so, an error will be reported.
error log is as follows:
2018-04-16 1515 CST, 55lane 45.450606 CST, "gp", "gp", p180444 Mathel 2040876928, "192.168.1.8", "25975", 2018-04-15 14:23:56 CST,376948,con289,cmd1,seg31,dx558714,x376948,sx1, "ERROR", "53100", "could not extend relation 1663384 could not extend relation 21258: No space left on device", "Check free disk space.", "create index index_table_name_bitmap_package on table_name USING btree (package); Smgr.c, 1335, "Stack trace:
1 0xae7486 postgres errstart (elog.c:503)
2 0x9ea06c postgres smgrextend (smgr.c:1329)
3 0x6db826 postgres < symbol not found > (nbtsort.c:338)
4 0x6dbc1a postgres < symbol not found > (nbtsort.c:602)
5 0x6dc7bf postgres _ bt_leafbuild (nbtsort.c:838)
6 0x6d716a postgres btbuild (nbtree.c:146)
7 0xaef042 postgres OidFunctionCall3 (fmgr.c:1595)
8 0x76a270 postgres index_build (index.c:1451)
9 0x76afdd postgres index_create (index.c)
10 0x8037c6 postgres DefineIndex (indexcmds.c:731)
11 0x9f5bdb postgres ProcessUtility (utility.c:1312)
12 0x9f2123 postgres < symbol not found > (pquery.c:1521)
13 0x9f3836 postgres < symbol not found > (pquery.c:1631)
14 0x9f49d5 postgres PortalRun (pquery.c:1129)
15 0x67c16e postgres < symbol not found > (discriminator 13)
16 0x9f0386 postgres PostgresMain (postgres.c:4819)
17 0x99776e postgres < symbol not found > (postmaster.c:6672)
18 0x999400 postgres PostmasterMain (postmaster.c:1559)
19 0x8d286b postgres C_main (main.c:206)
20 0x7f53837aeb35 libc.so.6 _ libc_start_main (?: 0)
21 0x69dcc1 postgres < symbol not found > (?

I made sure there was enough space, 20T space, 5 machines, each with 10 tables segment, and then left 8T before creating the index. Do you know why? Thank you

Mar.04,2021

use monitoring tools or the command line to monitor the history of disk usage changes.

creating an index uses up space, which is freed after an error is reported. If you check the space utilization rate after a mistake, you can't see it.

MySQL Query : SELECT * FROM `codeshelper`.`v9_news` WHERE status=99 AND catid='6' ORDER BY rand() LIMIT 5
MySQL Error : Disk full (/tmp/#sql-temptable-64f5-1b36eb9-2c038.MAI); waiting for someone to free some space... (errno: 28 "No space left on device")
MySQL Errno : 1021
Message : Disk full (/tmp/#sql-temptable-64f5-1b36eb9-2c038.MAI); waiting for someone to free some space... (errno: 28 "No space left on device")
Need Help?