Lili Ma created HAWQ-624:
----------------------------
Summary: Copy a large table to output file, multiple QEs are
started, but only one QE is assigned actual task.
Key: HAWQ-624
URL: https://issues.apache.org/jira/browse/HAWQ-624
Project: Apache HAWQ
Issue Type: Bug
Components: Core
Reporter: Lili Ma
Assignee: Lei Chang
Create a large table a, for example, with 1,000,000,000 records.
Then run copy command "copy a to '/tmp/a'".
Six QEs are started, but only one QE is assigned a valid split to scan, other
QEs scan split range is NULL.
Attach to one running QE, see the split information for each QE.
{code}
(lldb) bt
* thread #1: tid = 0x6a7b98, 0x00000001016bcdd6
postgres`appendonly_getnext(scan=0x00007fe5c4046430,
direction=ForwardScanDirection, slot=0x00007fe5c5000c40) + 54 at
appendonlyam.c:1614, queue = 'com.apple.main-thread', stop reason = step over
frame #0: 0x00000001016bcdd6
postgres`appendonly_getnext(scan=0x00007fe5c4046430,
direction=ForwardScanDirection, slot=0x00007fe5c5000c40) + 54 at
appendonlyam.c:1614
* frame #1: 0x00000001017f60ec postgres`CopyTo(cstate=0x00007fe5c6802030) +
2060 at copy.c:2360
frame #2: 0x00000001017ee4dc postgres`DoCopyTo(cstate=0x00007fe5c6802030) +
1452 at copy.c:1905
frame #3: 0x00000001017e4014 postgres`DoCopy(stmt=0x00007fe5c5824758,
queryString=0x00007fe5c58230f2) + 10676 at copy.c:1686
frame #4: 0x0000000101a992eb
postgres`ProcessUtility(parsetree=0x00007fe5c5824758,
queryString=0x00007fe5c7001c30, params=0x0000000000000000, isTopLevel='\x01',
dest=0x00007fe5c4037da0, completionTag=0x00007fff5e619bb0) + 3467 at
utility.c:1076
frame #5: 0x0000000101a97f53
postgres`PortalRunUtility(portal=0x00007fe5c5840230,
utilityStmt=0x00007fe5c5824758, isTopLevel='\x01', dest=0x00007fe5c4037da0,
completionTag=0x00007fff5e619bb0) + 467 at pquery.c:1969
frame #6: 0x0000000101a964a0
postgres`PortalRunMulti(portal=0x00007fe5c5840230, isTopLevel='\x01',
dest=0x00007fe5c4037da0, altdest=0x00007fe5c4037da0,
completionTag=0x00007fff5e619bb0) + 544 at pquery.c:2079
frame #7: 0x0000000101a959fb postgres`PortalRun(portal=0x00007fe5c5840230,
count=9223372036854775807, isTopLevel='\x01', dest=0x00007fe5c4037da0,
altdest=0x00007fe5c4037da0, completionTag=0x00007fff5e619bb0) + 1291 at
pquery.c:1596
frame #8: 0x0000000101a8cfc9
postgres`exec_mpp_query(query_string=0x00007fe5c58230f2,
serializedQuerytree=0x00007fe5c5823133, serializedQuerytreelen=980,
serializedPlantree=0x0000000000000000, serializedPlantreelen=0,
serializedParams=0x0000000000000000, serializedParamslen=0,
serializedSliceInfo=0x0000000000000000, serializedSliceInfolen=0,
serializedResource=0x00007fe5c5823554, serializedResourceLen=41,
seqServerHost=0x00007fe5c582357d, seqServerPort=54307, localSlice=0) + 5049 at
postgres.c:1414
frame #9: 0x0000000101a8a4e6 postgres`PostgresMain(argc=250,
argv=0x00007fe5c4826e30, username=0x00007fe5c4801670) + 9686 at postgres.c:4945
frame #10: 0x0000000101a2d1cb postgres`BackendRun(port=0x00007fe5c3c19360)
+ 1019 at postmaster.c:5889
frame #11: 0x0000000101a2c2a2
postgres`BackendStartup(port=0x00007fe5c3c19360) + 402 at postmaster.c:5484
frame #12: 0x0000000101a28d94 postgres`ServerLoop + 1348 at
postmaster.c:2163
frame #13: 0x0000000101a27350 postgres`PostmasterMain(argc=9,
argv=0x00007fe5c3c1d300) + 5072 at postmaster.c:1454
frame #14: 0x000000010192c211 postgres`main(argc=9,
argv=0x00007fe5c3c1d300) + 993 at main.c:226
frame #15: 0x00007fff8642e5c9 libdyld.dylib`start + 1
(lldb) p *cstate->splits
(List) $43 = {
type = T_List
length = 1
head = 0x00007fe5c402fc18
tail = 0x00007fe5c402fc18
}
(lldb) p *(ListCell*)0x00007fe5c402fc18
(ListCell) $44 = {
data = (ptr_value = void * = 0x00007fe5c402fa88, int_value = -1006437752,
oid_value = 3288529544)
next = 0x0000000000000000
}
(lldb) p *(SegFileSplitMapNode *)0x00007fe5c402fa88
(SegFileSplitMapNode) $45 = {
type = T_SegFileSplitMapNode
relid = 16508
splits = 0x00007fe5c402fb48
}
(lldb) p *(List*)0x00007fe5c402fb48
(List) $46 = {
type = T_List
length = 6
head = 0x00007fe5c402fb28
tail = 0x00007fe5c402fbf8
}
(lldb) p *(ListCell*)0x00007fe5c402fb28
(ListCell) $47 = {
data = (ptr_value = void * = 0x00007fe5c402faf8, int_value = -1006437640,
oid_value = 3288529656)
next = 0x00007fe5c402fb78
}
(lldb) p *(ListCell*)0x00007fe5c402fb78
(ListCell) $48 = {
data = (ptr_value = void * = 0x0000000000000000, int_value = 0, oid_value = 0)
next = 0x00007fe5c402fb98
}
(lldb) p *(ListCell*)0x00007fe5c402fb98
(ListCell) $49 = {
data = (ptr_value = void * = 0x0000000000000000, int_value = 0, oid_value = 0)
next = 0x00007fe5c402fbb8
}
(lldb) p *(ListCell*)0x00007fe5c402fbb8
(ListCell) $50 = {
data = (ptr_value = void * = 0x0000000000000000, int_value = 0, oid_value = 0)
next = 0x00007fe5c402fbd8
}
(lldb) p *(ListCell*)0x00007fe5c402fbd8
(ListCell) $51 = {
data = (ptr_value = void * = 0x0000000000000000, int_value = 0, oid_value = 0)
next = 0x00007fe5c402fbf8
}
(lldb) p *(ListCell*)0x00007fe5c402fbf8
(ListCell) $52 = {
data = (ptr_value = void * = 0x0000000000000000, int_value = 0, oid_value = 0)
next = 0x0000000000000000
}
{code}
Only the first listCell have valid split, the other 5 listCell are all NULL.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
