Skip to content
Projects
Groups
Snippets
Help
Loading...
Help
Support
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
N
neoppod
Project overview
Project overview
Details
Activity
Releases
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Issues
0
Issues
0
List
Boards
Labels
Milestones
Merge Requests
0
Merge Requests
0
CI / CD
CI / CD
Pipelines
Jobs
Schedules
Analytics
Analytics
CI / CD
Repository
Value Stream
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Create a new issue
Jobs
Commits
Issue Boards
Open sidebar
Levin Zimmermann
neoppod
Commits
69fd5651
Commit
69fd5651
authored
Oct 27, 2020
by
Kirill Smelkov
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
.
parent
eae3b427
Changes
2
Show whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
98 additions
and
105 deletions
+98
-105
go/neo/client.go
go/neo/client.go
+97
-104
go/zodb/storage/zeo/zeo.go
go/zodb/storage/zeo/zeo.go
+1
-1
No files found.
go/neo/client.go
View file @
69fd5651
...
...
@@ -31,6 +31,7 @@ import (
"github.com/pkg/errors"
"golang.org/x/sync/errgroup"
"lab.nexedi.com/kirr/go123/mem"
"lab.nexedi.com/kirr/go123/xerr"
"lab.nexedi.com/kirr/go123/xnet"
"lab.nexedi.com/kirr/neo/go/internal/xzlib"
...
...
@@ -77,6 +78,7 @@ type Client struct {
at0
zodb
.
Tid
// at0 obtained when initially connecting to server
eventq0
[]
*
zodb
.
EventCommit
// buffer for initial messages, until .at0 is initialized
at0Initialized
bool
// true after .at0 is initialized
at0Ready
chan
(
struct
{})
// ready after .at0 is initialized
}
var
_
zodb
.
IStorageDriver
=
(
*
Client
)(
nil
)
...
...
@@ -108,9 +110,6 @@ func (c *Client) Close() error {
c
.
talkMasterCancel
()
// XXX wait talkMaster finishes -> XXX return err from that?
// XXX what else?
if
c
.
watchq
!=
nil
{
close
(
c
.
watchq
)
}
return
nil
}
...
...
@@ -222,6 +221,8 @@ func (c *Client) talkMaster(ctx context.Context) (err error) {
// XXX dup wrt Storage.talkMaster
for
{
// XXX .nodeTab.Reset() ?
err
:=
c
.
talkMaster1
(
ctx
)
log
.
Warning
(
ctx
,
err
)
// XXX Warning ok? -> Error?
// XXX if err == "reject identification / protocol error" -> shutdown client
...
...
@@ -289,12 +290,68 @@ func (c *Client) talkMaster1(ctx context.Context) (err error) {
return
wg
.
Wait
()
}
// initFromMaster asks M for partTab and DB head right after identification.
func
(
c
*
Client
)
initFromMaster
(
ctx
context
.
Context
,
mlink
*
neonet
.
NodeLink
)
(
err
error
)
{
defer
task
.
Running
(
&
ctx
,
"init"
)(
&
err
)
// query partTab
rpt
:=
proto
.
AnswerPartitionTable
{}
err
=
mlink
.
Ask1
(
&
proto
.
AskPartitionTable
{},
&
rpt
)
if
err
!=
nil
{
return
err
}
pt
:=
PartTabFromDump
(
rpt
.
PTid
,
rpt
.
RowList
)
log
.
Infof
(
ctx
,
"master initialized us with next parttab:
\n
%s"
,
pt
)
c
.
node
.
StateMu
.
Lock
()
c
.
node
.
PartTab
=
pt
opready
:=
c
.
updateOperational
()
c
.
node
.
StateMu
.
Unlock
()
opready
()
// query last_tid
lastTxn
:=
proto
.
AnswerLastTransaction
{}
err
=
mlink
.
Ask1
(
&
proto
.
LastTransaction
{},
&
lastTxn
)
if
err
!=
nil
{
return
err
}
if
c
.
at0Initialized
{
// XXX c.head locking?
if
lastTxn
.
Tid
!=
c
.
head
{
return
fmt
.
Errorf
(
"New transactions were committed while we were disconnected from master (%s -> %s)"
,
c
.
head
,
lastTxn
.
Tid
)
}
}
// XXX -> else?
// since we read lastTid, in separate protocol exchange there is a
// chance, that by the time when lastTid was read, some new transactions
// were committed. This way lastTid will be > than some first
// transactions received by watcher via "invalidateObjects" server
// notification.
//
// filter-out first < at0 messages for this reason.
//
// TODO change NEO protocol so that when C connects to M, M sends it
// current head and guarantees to send only followup invalidation
// updates.
c
.
at0Mu
.
Lock
()
c
.
at0
=
lastTxn
.
Tid
c
.
at0Initialized
=
true
// c.flushEventq0() XXX reenable
c
.
at0Mu
.
Unlock
()
// XXX what next?
return
nil
// TODO transaction control? -> better in original goroutines doing the txn (just share mlink)
}
// recvMaster receives and handles notifications from master
func
(
c
*
Client
)
recvMaster
(
ctx
context
.
Context
,
mlink
*
neonet
.
NodeLink
)
(
err
error
)
{
defer
task
.
Running
(
&
ctx
,
"rx"
)(
&
err
)
// XXX .nodeTab.Reset()
for
{
req
,
err
:=
mlink
.
Recv1
()
if
err
!=
nil
{
...
...
@@ -345,61 +402,6 @@ func (c *Client) recvMaster1(ctx context.Context, req neonet.Request) error {
return
nil
}
func
(
c
*
Client
)
initFromMaster
(
ctx
context
.
Context
,
mlink
*
neonet
.
NodeLink
)
(
err
error
)
{
defer
task
.
Running
(
&
ctx
,
"init"
)(
&
err
)
// ask M for PT
rpt
:=
proto
.
AnswerPartitionTable
{}
err
=
mlink
.
Ask1
(
&
proto
.
AskPartitionTable
{},
&
rpt
)
if
err
!=
nil
{
return
err
}
pt
:=
PartTabFromDump
(
rpt
.
PTid
,
rpt
.
RowList
)
log
.
Infof
(
ctx
,
"master initialized us with next parttab:
\n
%s"
,
pt
)
c
.
node
.
StateMu
.
Lock
()
c
.
node
.
PartTab
=
pt
opready
:=
c
.
updateOperational
()
c
.
node
.
StateMu
.
Unlock
()
opready
()
// ask M about last_tid
lastTxn
:=
proto
.
AnswerLastTransaction
{}
err
=
mlink
.
Ask1
(
&
proto
.
LastTransaction
{},
&
lastTxn
)
if
err
!=
nil
{
return
err
}
if
c
.
at0Initialized
{
// XXX c.head locking?
if
lastTxn
.
Tid
!=
c
.
head
{
return
fmt
.
Errorf
(
"New transactions were committed while we were disconnected from master (%s -> %s)"
,
c
.
head
,
lastTxn
.
Tid
)
}
}
// XXX -> else?
// since we read lastTid, in separate protocol exchange there is a
// chance, that by the time when lastTid was read, some new transactions
// were committed. This way lastTid will be > than some first
// transactions received by watcher via "invalidateObjects" server
// notification.
//
// filter-out first < at0 messages for this reason.
//
// TODO change NEO protocol so that when C connects to M, M sends it
// current head and guarantees to send only followup invalidation
// updates.
c
.
at0Mu
.
Lock
()
c
.
at0
=
lastTxn
.
Tid
c
.
at0Initialized
=
true
// c.flushEventq0() XXX reenable
c
.
at0Mu
.
Unlock
()
// XXX what next?
return
nil
// TODO transaction control? -> better in original goroutines doing the txn (just share mlink)
}
// --- user API calls ---
func
(
c
*
Client
)
Sync
(
ctx
context
.
Context
)
(
_
zodb
.
Tid
,
err
error
)
{
...
...
@@ -530,67 +532,58 @@ func (c *Client) Watch(ctx context.Context) (zodb.Tid, []zodb.Oid, error) {
// ---- ZODB open/url support ----
func
openClientByURL
(
ctx
context
.
Context
,
u
*
url
.
URL
,
opt
*
zodb
.
DriverOptions
)
(
zodb
.
IStorageDriver
,
zodb
.
Tid
,
error
)
{
func
openClientByURL
(
ctx
context
.
Context
,
u
*
url
.
URL
,
opt
*
zodb
.
DriverOptions
)
(
_
zodb
.
IStorageDriver
,
_
zodb
.
Tid
,
err
error
)
{
// neo://name@master1,master2,...,masterN?options
defer
xerr
.
Contextf
(
&
err
,
"neo: open %s"
,
u
)
if
u
.
User
==
nil
{
return
nil
,
zodb
.
InvalidTid
,
fmt
.
Errorf
(
"
neo: open %q: cluster name not specified"
,
u
)
return
nil
,
zodb
.
InvalidTid
,
fmt
.
Errorf
(
"
cluster name not specified"
)
}
// XXX readonly stub
// XXX place = ?
if
!
opt
.
ReadOnly
{
return
nil
,
zodb
.
InvalidTid
,
fmt
.
Errorf
(
"neo: %s: TODO write mode not implemented"
,
u
)
}
// FIXME handle opt.Watchq
// for now we pretend as if the database is not changing.
// TODO watcher(when implementing): filter-out first < at0 messages.
if
opt
.
Watchq
!=
nil
{
log
.
Error
(
ctx
,
"neo: FIXME: watchq support not implemented - there"
+
"won't be notifications about database changes"
)
return
nil
,
zodb
.
InvalidTid
,
fmt
.
Errorf
(
"TODO write mode not implemented"
)
}
// XXX check/use other url fields
net
:=
xnet
.
NetPlain
(
"tcp"
)
// TODO + TLS; not only "tcp" ?
// XXX we are not passing ctx to NewClient - right?
// as ctx for open can be done after open finishes - not covering
// whole storage working lifetime.
c
:=
NewClient
(
u
.
User
.
Username
(),
u
.
Host
,
net
)
c
.
watchq
=
opt
.
Watchq
defer
func
()
{
if
err
!=
nil
{
c
.
Close
()
}
}()
// XXX go c.Run()
// XXX wait c.ready
// start client node serve loop
errq
:=
make
(
chan
error
,
1
)
go
func
()
{
err
:=
c
.
Run
(
context
.
Background
())
// NOTE not ctx
/*
lastTid, err := c.Sync(ctx)
// close .watchq after serve is over
c
.
at0Mu
.
Lock
()
defer
c
.
at0Mu
.
Unlock
()
if
c
.
at0Initialized
{
// c.flushEventq0()
}
if
c
.
watchq
!=
nil
{
if
err
!=
nil
{
c.Close() // XXX lclose
return nil, zodb.InvalidTid, fmt.Errorf("neo: open %q: %s", u, err)
c
.
watchq
<-
&
zodb
.
EventError
{
Err
:
err
}
}
close
(
c
.
watchq
)
}
// since we read lastTid, in separate protocol exchange there is a
// chance, that by the time when lastTid was read, some new transactions
// were committed. This way lastTid will be > than some first
// transactions received by watcher via "invalidateObjects" server
// notification.
//
// filter-out first < at0 messages for this reason.
//
// TODO change NEO protocol so that when C connects to M, M sends it
// current head and guarantees to send only followup invalidation
// updates.
//
// XXX -> move talkMaster1 / Client.Run
c.at0Mu.Lock()
c.at0 = lastTid
c.at0Initialized = true
c.flushEventq0()
c.at0Mu.Unlock()
*/
errq
<-
err
}()
select
{
case
<-
ctx
.
Done
()
:
return
nil
,
zodb
.
InvalidTid
,
ctx
.
Err
()
case
<-
errq
:
return
nil
,
zodb
.
InvalidTid
,
err
case
<-
c
.
at0Ready
:
return
c
,
c
.
at0
,
nil
}
}
func
(
c
*
Client
)
URL
()
string
{
...
...
go/zodb/storage/zeo/zeo.go
View file @
69fd5651
...
...
@@ -390,7 +390,7 @@ func (r rpc) ereplyf(format string, argv ...interface{}) *errorUnexpectedReply {
func
openByURL
(
ctx
context
.
Context
,
u
*
url
.
URL
,
opt
*
zodb
.
DriverOptions
)
(
_
zodb
.
IStorageDriver
,
at0
zodb
.
Tid
,
err
error
)
{
url
:=
u
.
String
()
defer
xerr
.
Contextf
(
&
err
,
"open %s"
,
url
)
defer
xerr
.
Contextf
(
&
err
,
"
zeo:
open %s"
,
url
)
// zeo://host:port/path?storage=...&...
var
net
xnet
.
Networker
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment